Merge remote-tracking branch 'origin/main' into optimal-layout

This commit is contained in:
Mendes
2022-10-04 18:14:49 +02:00
260 changed files with 30550 additions and 5813 deletions
-29
View File
@@ -1,29 +0,0 @@
[package]
name = "garage_admin"
version = "0.7.0"
authors = ["Maximilien Richer <code@mricher.fr>"]
edition = "2018"
license = "AGPL-3.0"
description = "Administration and metrics REST HTTP server for Garage"
repository = "https://git.deuxfleurs.fr/Deuxfleurs/garage"
[lib]
path = "lib.rs"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
garage_util = { version = "0.7.0", path = "../util" }
hex = "0.4"
futures = "0.3"
futures-util = "0.3"
http = "0.2"
hyper = "0.14"
tracing = "0.1.30"
opentelemetry = { version = "0.17", features = [ "rt-tokio" ] }
opentelemetry-prometheus = "0.10"
opentelemetry-otlp = "0.10"
prometheus = "0.13"
-6
View File
@@ -1,6 +0,0 @@
//! Crate for handling the admin and metric HTTP APIs
#[macro_use]
extern crate tracing;
pub mod metrics;
pub mod tracing_setup;
-146
View File
@@ -1,146 +0,0 @@
use std::convert::Infallible;
use std::net::SocketAddr;
use std::sync::Arc;
use std::time::SystemTime;
use futures::future::*;
use hyper::{
header::CONTENT_TYPE,
service::{make_service_fn, service_fn},
Body, Method, Request, Response, Server,
};
use opentelemetry::{
global,
metrics::{BoundCounter, BoundValueRecorder},
trace::{FutureExt, TraceContextExt, Tracer},
Context,
};
use opentelemetry_prometheus::PrometheusExporter;
use prometheus::{Encoder, TextEncoder};
use garage_util::error::Error as GarageError;
use garage_util::metrics::*;
// serve_req on metric endpoint
async fn serve_req(
req: Request<Body>,
admin_server: Arc<AdminServer>,
) -> Result<Response<Body>, hyper::Error> {
debug!("Receiving request at path {}", req.uri());
let request_start = SystemTime::now();
admin_server.metrics.http_counter.add(1);
let response = match (req.method(), req.uri().path()) {
(&Method::GET, "/metrics") => {
let mut buffer = vec![];
let encoder = TextEncoder::new();
let tracer = opentelemetry::global::tracer("garage");
let metric_families = tracer.in_span("admin/gather_metrics", |_| {
admin_server.exporter.registry().gather()
});
encoder.encode(&metric_families, &mut buffer).unwrap();
admin_server
.metrics
.http_body_gauge
.record(buffer.len() as u64);
Response::builder()
.status(200)
.header(CONTENT_TYPE, encoder.format_type())
.body(Body::from(buffer))
.unwrap()
}
_ => Response::builder()
.status(404)
.body(Body::from("Not implemented"))
.unwrap(),
};
admin_server
.metrics
.http_req_histogram
.record(request_start.elapsed().map_or(0.0, |d| d.as_secs_f64()));
Ok(response)
}
// AdminServer hold the admin server internal admin_server and the metric exporter
pub struct AdminServer {
exporter: PrometheusExporter,
metrics: AdminServerMetrics,
}
// GarageMetricadmin_server holds the metrics counter definition for Garage
// FIXME: we would rather have that split up among the different libraries?
struct AdminServerMetrics {
http_counter: BoundCounter<u64>,
http_body_gauge: BoundValueRecorder<u64>,
http_req_histogram: BoundValueRecorder<f64>,
}
impl AdminServer {
/// init initilialize the AdminServer and background metric server
pub fn init() -> AdminServer {
let exporter = opentelemetry_prometheus::exporter().init();
let meter = global::meter("garage/admin_server");
AdminServer {
exporter,
metrics: AdminServerMetrics {
http_counter: meter
.u64_counter("admin.http_requests_total")
.with_description("Total number of HTTP requests made.")
.init()
.bind(&[]),
http_body_gauge: meter
.u64_value_recorder("admin.http_response_size_bytes")
.with_description("The metrics HTTP response sizes in bytes.")
.init()
.bind(&[]),
http_req_histogram: meter
.f64_value_recorder("admin.http_request_duration_seconds")
.with_description("The HTTP request latencies in seconds.")
.init()
.bind(&[]),
},
}
}
/// run execute the admin server on the designated HTTP port and listen for requests
pub async fn run(
self,
bind_addr: SocketAddr,
shutdown_signal: impl Future<Output = ()>,
) -> Result<(), GarageError> {
let admin_server = Arc::new(self);
// For every connection, we must make a `Service` to handle all
// incoming HTTP requests on said connection.
let make_svc = make_service_fn(move |_conn| {
let admin_server = admin_server.clone();
// This is the `Service` that will handle the connection.
// `service_fn` is a helper to convert a function that
// returns a Response into a `Service`.
async move {
Ok::<_, Infallible>(service_fn(move |req| {
let tracer = opentelemetry::global::tracer("garage");
let span = tracer
.span_builder("admin/request")
.with_trace_id(gen_trace_id())
.start(&tracer);
serve_req(req, admin_server.clone())
.with_context(Context::current_with_span(span))
}))
}
});
let server = Server::bind(&bind_addr).serve(make_svc);
let graceful = server.with_graceful_shutdown(shutdown_signal);
info!("Admin server listening on http://{}", bind_addr);
graceful.await?;
Ok(())
}
}
+18 -9
View File
@@ -1,6 +1,6 @@
[package]
name = "garage_api"
version = "0.7.0"
version = "0.8.0"
authors = ["Alex Auvolat <alex@adnab.me>"]
edition = "2018"
license = "AGPL-3.0"
@@ -14,28 +14,31 @@ path = "lib.rs"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
garage_model = { version = "0.7.0", path = "../model" }
garage_table = { version = "0.7.0", path = "../table" }
garage_block = { version = "0.7.0", path = "../block" }
garage_util = { version = "0.7.0", path = "../util" }
garage_model = { version = "0.8.0", path = "../model" }
garage_table = { version = "0.8.0", path = "../table" }
garage_block = { version = "0.8.0", path = "../block" }
garage_util = { version = "0.8.0", path = "../util" }
garage_rpc = { version = "0.8.0", path = "../rpc" }
async-trait = "0.1.7"
base64 = "0.13"
bytes = "1.0"
chrono = "0.4"
crypto-mac = "0.10"
crypto-common = "0.1"
err-derive = "0.3"
hex = "0.4"
hmac = "0.10"
hmac = "0.12"
idna = "0.2"
tracing = "0.1.30"
md-5 = "0.9"
md-5 = "0.10"
nom = "7.1"
sha2 = "0.9"
sha2 = "0.10"
futures = "0.3"
futures-util = "0.3"
pin-project = "1.0"
tokio = { version = "1.0", default-features = false, features = ["rt", "rt-multi-thread", "io-util", "net", "time", "macros", "sync", "signal", "fs"] }
tokio-stream = "0.1"
form_urlencoded = "1.0.0"
http = "0.2"
@@ -52,3 +55,9 @@ quick-xml = { version = "0.21", features = [ "serialize" ] }
url = "2.1"
opentelemetry = "0.17"
opentelemetry-prometheus = { version = "0.10", optional = true }
prometheus = { version = "0.13", optional = true }
[features]
k2v = [ "garage_util/k2v", "garage_model/k2v" ]
metrics = [ "opentelemetry-prometheus", "prometheus" ]
+209
View File
@@ -0,0 +1,209 @@
use std::net::SocketAddr;
use std::sync::Arc;
use async_trait::async_trait;
use futures::future::Future;
use http::header::{ACCESS_CONTROL_ALLOW_METHODS, ACCESS_CONTROL_ALLOW_ORIGIN, ALLOW};
use hyper::{Body, Request, Response};
use opentelemetry::trace::SpanRef;
#[cfg(feature = "metrics")]
use opentelemetry_prometheus::PrometheusExporter;
#[cfg(feature = "metrics")]
use prometheus::{Encoder, TextEncoder};
use garage_model::garage::Garage;
use garage_util::error::Error as GarageError;
use crate::generic_server::*;
use crate::admin::bucket::*;
use crate::admin::cluster::*;
use crate::admin::error::*;
use crate::admin::key::*;
use crate::admin::router::{Authorization, Endpoint};
pub struct AdminApiServer {
garage: Arc<Garage>,
#[cfg(feature = "metrics")]
exporter: PrometheusExporter,
metrics_token: Option<String>,
admin_token: Option<String>,
}
impl AdminApiServer {
pub fn new(
garage: Arc<Garage>,
#[cfg(feature = "metrics")] exporter: PrometheusExporter,
) -> Self {
let cfg = &garage.config.admin;
let metrics_token = cfg
.metrics_token
.as_ref()
.map(|tok| format!("Bearer {}", tok));
let admin_token = cfg
.admin_token
.as_ref()
.map(|tok| format!("Bearer {}", tok));
Self {
garage,
#[cfg(feature = "metrics")]
exporter,
metrics_token,
admin_token,
}
}
pub async fn run(
self,
bind_addr: SocketAddr,
shutdown_signal: impl Future<Output = ()>,
) -> Result<(), GarageError> {
let region = self.garage.config.s3_api.s3_region.clone();
ApiServer::new(region, self)
.run_server(bind_addr, shutdown_signal)
.await
}
fn handle_options(&self, _req: &Request<Body>) -> Result<Response<Body>, Error> {
Ok(Response::builder()
.status(204)
.header(ALLOW, "OPTIONS, GET, POST")
.header(ACCESS_CONTROL_ALLOW_METHODS, "OPTIONS, GET, POST")
.header(ACCESS_CONTROL_ALLOW_ORIGIN, "*")
.body(Body::empty())?)
}
fn handle_metrics(&self) -> Result<Response<Body>, Error> {
#[cfg(feature = "metrics")]
{
use opentelemetry::trace::Tracer;
let mut buffer = vec![];
let encoder = TextEncoder::new();
let tracer = opentelemetry::global::tracer("garage");
let metric_families = tracer.in_span("admin/gather_metrics", |_| {
self.exporter.registry().gather()
});
encoder
.encode(&metric_families, &mut buffer)
.ok_or_internal_error("Could not serialize metrics")?;
Ok(Response::builder()
.status(200)
.header(http::header::CONTENT_TYPE, encoder.format_type())
.body(Body::from(buffer))?)
}
#[cfg(not(feature = "metrics"))]
Err(Error::bad_request(
"Garage was built without the metrics feature".to_string(),
))
}
}
#[async_trait]
impl ApiHandler for AdminApiServer {
const API_NAME: &'static str = "admin";
const API_NAME_DISPLAY: &'static str = "Admin";
type Endpoint = Endpoint;
type Error = Error;
fn parse_endpoint(&self, req: &Request<Body>) -> Result<Endpoint, Error> {
Endpoint::from_request(req)
}
async fn handle(
&self,
req: Request<Body>,
endpoint: Endpoint,
) -> Result<Response<Body>, Error> {
let expected_auth_header =
match endpoint.authorization_type() {
Authorization::MetricsToken => self.metrics_token.as_ref(),
Authorization::AdminToken => match &self.admin_token {
None => return Err(Error::forbidden(
"Admin token isn't configured, admin API access is disabled for security.",
)),
Some(t) => Some(t),
},
};
if let Some(h) = expected_auth_header {
match req.headers().get("Authorization") {
None => return Err(Error::forbidden("Authorization token must be provided")),
Some(v) => {
let authorized = v.to_str().map(|hv| hv.trim() == h).unwrap_or(false);
if !authorized {
return Err(Error::forbidden("Invalid authorization token provided"));
}
}
}
}
match endpoint {
Endpoint::Options => self.handle_options(&req),
Endpoint::Metrics => self.handle_metrics(),
Endpoint::GetClusterStatus => handle_get_cluster_status(&self.garage).await,
Endpoint::ConnectClusterNodes => handle_connect_cluster_nodes(&self.garage, req).await,
// Layout
Endpoint::GetClusterLayout => handle_get_cluster_layout(&self.garage).await,
Endpoint::UpdateClusterLayout => handle_update_cluster_layout(&self.garage, req).await,
Endpoint::ApplyClusterLayout => handle_apply_cluster_layout(&self.garage, req).await,
Endpoint::RevertClusterLayout => handle_revert_cluster_layout(&self.garage, req).await,
// Keys
Endpoint::ListKeys => handle_list_keys(&self.garage).await,
Endpoint::GetKeyInfo { id, search } => {
handle_get_key_info(&self.garage, id, search).await
}
Endpoint::CreateKey => handle_create_key(&self.garage, req).await,
Endpoint::ImportKey => handle_import_key(&self.garage, req).await,
Endpoint::UpdateKey { id } => handle_update_key(&self.garage, id, req).await,
Endpoint::DeleteKey { id } => handle_delete_key(&self.garage, id).await,
// Buckets
Endpoint::ListBuckets => handle_list_buckets(&self.garage).await,
Endpoint::GetBucketInfo { id, global_alias } => {
handle_get_bucket_info(&self.garage, id, global_alias).await
}
Endpoint::CreateBucket => handle_create_bucket(&self.garage, req).await,
Endpoint::DeleteBucket { id } => handle_delete_bucket(&self.garage, id).await,
Endpoint::UpdateBucket { id } => handle_update_bucket(&self.garage, id, req).await,
// Bucket-key permissions
Endpoint::BucketAllowKey => {
handle_bucket_change_key_perm(&self.garage, req, true).await
}
Endpoint::BucketDenyKey => {
handle_bucket_change_key_perm(&self.garage, req, false).await
}
// Bucket aliasing
Endpoint::GlobalAliasBucket { id, alias } => {
handle_global_alias_bucket(&self.garage, id, alias).await
}
Endpoint::GlobalUnaliasBucket { id, alias } => {
handle_global_unalias_bucket(&self.garage, id, alias).await
}
Endpoint::LocalAliasBucket {
id,
access_key_id,
alias,
} => handle_local_alias_bucket(&self.garage, id, access_key_id, alias).await,
Endpoint::LocalUnaliasBucket {
id,
access_key_id,
alias,
} => handle_local_unalias_bucket(&self.garage, id, access_key_id, alias).await,
}
}
}
impl ApiEndpoint for Endpoint {
fn name(&self) -> &'static str {
Endpoint::name(self)
}
fn add_span_attributes(&self, _span: SpanRef<'_>) {}
}
+580
View File
@@ -0,0 +1,580 @@
use std::collections::HashMap;
use std::sync::Arc;
use hyper::{Body, Request, Response, StatusCode};
use serde::{Deserialize, Serialize};
use garage_util::crdt::*;
use garage_util::data::*;
use garage_util::time::*;
use garage_table::*;
use garage_model::bucket_alias_table::*;
use garage_model::bucket_table::*;
use garage_model::garage::Garage;
use garage_model::permission::*;
use garage_model::s3::object_table::*;
use crate::admin::error::*;
use crate::admin::key::ApiBucketKeyPerm;
use crate::common_error::CommonError;
use crate::helpers::{json_ok_response, parse_json_body};
pub async fn handle_list_buckets(garage: &Arc<Garage>) -> Result<Response<Body>, Error> {
let buckets = garage
.bucket_table
.get_range(
&EmptyKey,
None,
Some(DeletedFilter::NotDeleted),
10000,
EnumerationOrder::Forward,
)
.await?;
let res = buckets
.into_iter()
.map(|b| {
let state = b.state.as_option().unwrap();
ListBucketResultItem {
id: hex::encode(b.id),
global_aliases: state
.aliases
.items()
.iter()
.filter(|(_, _, a)| *a)
.map(|(n, _, _)| n.to_string())
.collect::<Vec<_>>(),
local_aliases: state
.local_aliases
.items()
.iter()
.filter(|(_, _, a)| *a)
.map(|((k, n), _, _)| BucketLocalAlias {
access_key_id: k.to_string(),
alias: n.to_string(),
})
.collect::<Vec<_>>(),
}
})
.collect::<Vec<_>>();
Ok(json_ok_response(&res)?)
}
#[derive(Serialize)]
#[serde(rename_all = "camelCase")]
struct ListBucketResultItem {
id: String,
global_aliases: Vec<String>,
local_aliases: Vec<BucketLocalAlias>,
}
#[derive(Serialize)]
#[serde(rename_all = "camelCase")]
struct BucketLocalAlias {
access_key_id: String,
alias: String,
}
#[derive(Serialize, Deserialize)]
#[serde(rename_all = "camelCase")]
struct ApiBucketQuotas {
max_size: Option<u64>,
max_objects: Option<u64>,
}
pub async fn handle_get_bucket_info(
garage: &Arc<Garage>,
id: Option<String>,
global_alias: Option<String>,
) -> Result<Response<Body>, Error> {
let bucket_id = match (id, global_alias) {
(Some(id), None) => parse_bucket_id(&id)?,
(None, Some(ga)) => garage
.bucket_helper()
.resolve_global_bucket_name(&ga)
.await?
.ok_or_else(|| HelperError::NoSuchBucket(ga.to_string()))?,
_ => {
return Err(Error::bad_request(
"Either id or globalAlias must be provided (but not both)",
));
}
};
bucket_info_results(garage, bucket_id).await
}
async fn bucket_info_results(
garage: &Arc<Garage>,
bucket_id: Uuid,
) -> Result<Response<Body>, Error> {
let bucket = garage
.bucket_helper()
.get_existing_bucket(bucket_id)
.await?;
let counters = garage
.object_counter_table
.table
.get(&bucket_id, &EmptyKey)
.await?
.map(|x| x.filtered_values(&garage.system.ring.borrow()))
.unwrap_or_default();
let mut relevant_keys = HashMap::new();
for (k, _) in bucket
.state
.as_option()
.unwrap()
.authorized_keys
.items()
.iter()
{
if let Some(key) = garage
.key_table
.get(&EmptyKey, k)
.await?
.filter(|k| !k.is_deleted())
{
if !key.state.is_deleted() {
relevant_keys.insert(k.clone(), key);
}
}
}
for ((k, _), _, _) in bucket
.state
.as_option()
.unwrap()
.local_aliases
.items()
.iter()
{
if relevant_keys.contains_key(k) {
continue;
}
if let Some(key) = garage.key_table.get(&EmptyKey, k).await? {
if !key.state.is_deleted() {
relevant_keys.insert(k.clone(), key);
}
}
}
let state = bucket.state.as_option().unwrap();
let quotas = state.quotas.get();
let res =
GetBucketInfoResult {
id: hex::encode(&bucket.id),
global_aliases: state
.aliases
.items()
.iter()
.filter(|(_, _, a)| *a)
.map(|(n, _, _)| n.to_string())
.collect::<Vec<_>>(),
website_access: state.website_config.get().is_some(),
website_config: state.website_config.get().clone().map(|wsc| {
GetBucketInfoWebsiteResult {
index_document: wsc.index_document,
error_document: wsc.error_document,
}
}),
keys: relevant_keys
.into_iter()
.map(|(_, key)| {
let p = key.state.as_option().unwrap();
GetBucketInfoKey {
access_key_id: key.key_id,
name: p.name.get().to_string(),
permissions: p
.authorized_buckets
.get(&bucket.id)
.map(|p| ApiBucketKeyPerm {
read: p.allow_read,
write: p.allow_write,
owner: p.allow_owner,
})
.unwrap_or_default(),
bucket_local_aliases: p
.local_aliases
.items()
.iter()
.filter(|(_, _, b)| *b == Some(bucket.id))
.map(|(n, _, _)| n.to_string())
.collect::<Vec<_>>(),
}
})
.collect::<Vec<_>>(),
objects: counters.get(OBJECTS).cloned().unwrap_or_default(),
bytes: counters.get(BYTES).cloned().unwrap_or_default(),
unfinshed_uploads: counters
.get(UNFINISHED_UPLOADS)
.cloned()
.unwrap_or_default(),
quotas: ApiBucketQuotas {
max_size: quotas.max_size,
max_objects: quotas.max_objects,
},
};
Ok(json_ok_response(&res)?)
}
#[derive(Serialize)]
#[serde(rename_all = "camelCase")]
struct GetBucketInfoResult {
id: String,
global_aliases: Vec<String>,
website_access: bool,
#[serde(default)]
website_config: Option<GetBucketInfoWebsiteResult>,
keys: Vec<GetBucketInfoKey>,
objects: i64,
bytes: i64,
unfinshed_uploads: i64,
quotas: ApiBucketQuotas,
}
#[derive(Serialize)]
#[serde(rename_all = "camelCase")]
struct GetBucketInfoWebsiteResult {
index_document: String,
error_document: Option<String>,
}
#[derive(Serialize)]
#[serde(rename_all = "camelCase")]
struct GetBucketInfoKey {
access_key_id: String,
name: String,
permissions: ApiBucketKeyPerm,
bucket_local_aliases: Vec<String>,
}
pub async fn handle_create_bucket(
garage: &Arc<Garage>,
req: Request<Body>,
) -> Result<Response<Body>, Error> {
let req = parse_json_body::<CreateBucketRequest>(req).await?;
if let Some(ga) = &req.global_alias {
if !is_valid_bucket_name(ga) {
return Err(Error::bad_request(format!(
"{}: {}",
ga, INVALID_BUCKET_NAME_MESSAGE
)));
}
if let Some(alias) = garage.bucket_alias_table.get(&EmptyKey, ga).await? {
if alias.state.get().is_some() {
return Err(CommonError::BucketAlreadyExists.into());
}
}
}
if let Some(la) = &req.local_alias {
if !is_valid_bucket_name(&la.alias) {
return Err(Error::bad_request(format!(
"{}: {}",
la.alias, INVALID_BUCKET_NAME_MESSAGE
)));
}
let key = garage
.key_helper()
.get_existing_key(&la.access_key_id)
.await?;
let state = key.state.as_option().unwrap();
if matches!(state.local_aliases.get(&la.alias), Some(_)) {
return Err(Error::bad_request("Local alias already exists"));
}
}
let bucket = Bucket::new();
garage.bucket_table.insert(&bucket).await?;
if let Some(ga) = &req.global_alias {
garage
.bucket_helper()
.set_global_bucket_alias(bucket.id, ga)
.await?;
}
if let Some(la) = &req.local_alias {
garage
.bucket_helper()
.set_local_bucket_alias(bucket.id, &la.access_key_id, &la.alias)
.await?;
if la.allow.read || la.allow.write || la.allow.owner {
garage
.bucket_helper()
.set_bucket_key_permissions(
bucket.id,
&la.access_key_id,
BucketKeyPerm {
timestamp: now_msec(),
allow_read: la.allow.read,
allow_write: la.allow.write,
allow_owner: la.allow.owner,
},
)
.await?;
}
}
bucket_info_results(garage, bucket.id).await
}
#[derive(Deserialize)]
#[serde(rename_all = "camelCase")]
struct CreateBucketRequest {
global_alias: Option<String>,
local_alias: Option<CreateBucketLocalAlias>,
}
#[derive(Deserialize)]
#[serde(rename_all = "camelCase")]
struct CreateBucketLocalAlias {
access_key_id: String,
alias: String,
#[serde(default)]
allow: ApiBucketKeyPerm,
}
pub async fn handle_delete_bucket(
garage: &Arc<Garage>,
id: String,
) -> Result<Response<Body>, Error> {
let helper = garage.bucket_helper();
let bucket_id = parse_bucket_id(&id)?;
let mut bucket = helper.get_existing_bucket(bucket_id).await?;
let state = bucket.state.as_option().unwrap();
// Check bucket is empty
if !helper.is_bucket_empty(bucket_id).await? {
return Err(CommonError::BucketNotEmpty.into());
}
// --- done checking, now commit ---
// 1. delete authorization from keys that had access
for (key_id, perm) in bucket.authorized_keys() {
if perm.is_any() {
helper
.set_bucket_key_permissions(bucket.id, key_id, BucketKeyPerm::NO_PERMISSIONS)
.await?;
}
}
// 2. delete all local aliases
for ((key_id, alias), _, active) in state.local_aliases.items().iter() {
if *active {
helper
.unset_local_bucket_alias(bucket.id, key_id, alias)
.await?;
}
}
// 3. delete all global aliases
for (alias, _, active) in state.aliases.items().iter() {
if *active {
helper.purge_global_bucket_alias(bucket.id, alias).await?;
}
}
// 4. delete bucket
bucket.state = Deletable::delete();
garage.bucket_table.insert(&bucket).await?;
Ok(Response::builder()
.status(StatusCode::NO_CONTENT)
.body(Body::empty())?)
}
pub async fn handle_update_bucket(
garage: &Arc<Garage>,
id: String,
req: Request<Body>,
) -> Result<Response<Body>, Error> {
let req = parse_json_body::<UpdateBucketRequest>(req).await?;
let bucket_id = parse_bucket_id(&id)?;
let mut bucket = garage
.bucket_helper()
.get_existing_bucket(bucket_id)
.await?;
let state = bucket.state.as_option_mut().unwrap();
if let Some(wa) = req.website_access {
if wa.enabled {
state.website_config.update(Some(WebsiteConfig {
index_document: wa.index_document.ok_or_bad_request(
"Please specify indexDocument when enabling website access.",
)?,
error_document: wa.error_document,
}));
} else {
if wa.index_document.is_some() || wa.error_document.is_some() {
return Err(Error::bad_request(
"Cannot specify indexDocument or errorDocument when disabling website access.",
));
}
state.website_config.update(None);
}
}
if let Some(q) = req.quotas {
state.quotas.update(BucketQuotas {
max_size: q.max_size,
max_objects: q.max_objects,
});
}
garage.bucket_table.insert(&bucket).await?;
bucket_info_results(garage, bucket_id).await
}
#[derive(Deserialize)]
#[serde(rename_all = "camelCase")]
struct UpdateBucketRequest {
website_access: Option<UpdateBucketWebsiteAccess>,
quotas: Option<ApiBucketQuotas>,
}
#[derive(Deserialize)]
#[serde(rename_all = "camelCase")]
struct UpdateBucketWebsiteAccess {
enabled: bool,
index_document: Option<String>,
error_document: Option<String>,
}
// ---- BUCKET/KEY PERMISSIONS ----
pub async fn handle_bucket_change_key_perm(
garage: &Arc<Garage>,
req: Request<Body>,
new_perm_flag: bool,
) -> Result<Response<Body>, Error> {
let req = parse_json_body::<BucketKeyPermChangeRequest>(req).await?;
let bucket_id = parse_bucket_id(&req.bucket_id)?;
let bucket = garage
.bucket_helper()
.get_existing_bucket(bucket_id)
.await?;
let state = bucket.state.as_option().unwrap();
let key = garage
.key_helper()
.get_existing_key(&req.access_key_id)
.await?;
let mut perm = state
.authorized_keys
.get(&key.key_id)
.cloned()
.unwrap_or(BucketKeyPerm::NO_PERMISSIONS);
if req.permissions.read {
perm.allow_read = new_perm_flag;
}
if req.permissions.write {
perm.allow_write = new_perm_flag;
}
if req.permissions.owner {
perm.allow_owner = new_perm_flag;
}
garage
.bucket_helper()
.set_bucket_key_permissions(bucket.id, &key.key_id, perm)
.await?;
bucket_info_results(garage, bucket.id).await
}
#[derive(Deserialize)]
#[serde(rename_all = "camelCase")]
struct BucketKeyPermChangeRequest {
bucket_id: String,
access_key_id: String,
permissions: ApiBucketKeyPerm,
}
// ---- BUCKET ALIASES ----
pub async fn handle_global_alias_bucket(
garage: &Arc<Garage>,
bucket_id: String,
alias: String,
) -> Result<Response<Body>, Error> {
let bucket_id = parse_bucket_id(&bucket_id)?;
garage
.bucket_helper()
.set_global_bucket_alias(bucket_id, &alias)
.await?;
bucket_info_results(garage, bucket_id).await
}
pub async fn handle_global_unalias_bucket(
garage: &Arc<Garage>,
bucket_id: String,
alias: String,
) -> Result<Response<Body>, Error> {
let bucket_id = parse_bucket_id(&bucket_id)?;
garage
.bucket_helper()
.unset_global_bucket_alias(bucket_id, &alias)
.await?;
bucket_info_results(garage, bucket_id).await
}
pub async fn handle_local_alias_bucket(
garage: &Arc<Garage>,
bucket_id: String,
access_key_id: String,
alias: String,
) -> Result<Response<Body>, Error> {
let bucket_id = parse_bucket_id(&bucket_id)?;
garage
.bucket_helper()
.set_local_bucket_alias(bucket_id, &access_key_id, &alias)
.await?;
bucket_info_results(garage, bucket_id).await
}
pub async fn handle_local_unalias_bucket(
garage: &Arc<Garage>,
bucket_id: String,
access_key_id: String,
alias: String,
) -> Result<Response<Body>, Error> {
let bucket_id = parse_bucket_id(&bucket_id)?;
garage
.bucket_helper()
.unset_local_bucket_alias(bucket_id, &access_key_id, &alias)
.await?;
bucket_info_results(garage, bucket_id).await
}
// ---- HELPER ----
fn parse_bucket_id(id: &str) -> Result<Uuid, Error> {
let id_hex = hex::decode(&id).ok_or_bad_request("Invalid bucket id")?;
Ok(Uuid::try_from(&id_hex).ok_or_bad_request("Invalid bucket id")?)
}
+193
View File
@@ -0,0 +1,193 @@
use std::collections::HashMap;
use std::net::SocketAddr;
use std::sync::Arc;
use hyper::{Body, Request, Response, StatusCode};
use serde::{Deserialize, Serialize};
use garage_util::crdt::*;
use garage_util::data::*;
use garage_rpc::layout::*;
use garage_model::garage::Garage;
use crate::admin::error::*;
use crate::helpers::{json_ok_response, parse_json_body};
pub async fn handle_get_cluster_status(garage: &Arc<Garage>) -> Result<Response<Body>, Error> {
let res = GetClusterStatusResponse {
node: hex::encode(garage.system.id),
garage_version: garage_util::version::garage_version(),
garage_features: garage_util::version::garage_features(),
db_engine: garage.db.engine(),
known_nodes: garage
.system
.get_known_nodes()
.into_iter()
.map(|i| {
(
hex::encode(i.id),
KnownNodeResp {
addr: i.addr,
is_up: i.is_up,
last_seen_secs_ago: i.last_seen_secs_ago,
hostname: i.status.hostname,
},
)
})
.collect(),
layout: get_cluster_layout(garage),
};
Ok(json_ok_response(&res)?)
}
pub async fn handle_connect_cluster_nodes(
garage: &Arc<Garage>,
req: Request<Body>,
) -> Result<Response<Body>, Error> {
let req = parse_json_body::<Vec<String>>(req).await?;
let res = futures::future::join_all(req.iter().map(|node| garage.system.connect(node)))
.await
.into_iter()
.map(|r| match r {
Ok(()) => ConnectClusterNodesResponse {
success: true,
error: None,
},
Err(e) => ConnectClusterNodesResponse {
success: false,
error: Some(format!("{}", e)),
},
})
.collect::<Vec<_>>();
Ok(json_ok_response(&res)?)
}
pub async fn handle_get_cluster_layout(garage: &Arc<Garage>) -> Result<Response<Body>, Error> {
let res = get_cluster_layout(garage);
Ok(json_ok_response(&res)?)
}
fn get_cluster_layout(garage: &Arc<Garage>) -> GetClusterLayoutResponse {
let layout = garage.system.get_cluster_layout();
GetClusterLayoutResponse {
version: layout.version,
roles: layout
.roles
.items()
.iter()
.filter(|(_, _, v)| v.0.is_some())
.map(|(k, _, v)| (hex::encode(k), v.0.clone()))
.collect(),
staged_role_changes: layout
.staging
.items()
.iter()
.filter(|(k, _, v)| layout.roles.get(k) != Some(v))
.map(|(k, _, v)| (hex::encode(k), v.0.clone()))
.collect(),
}
}
#[derive(Serialize)]
#[serde(rename_all = "camelCase")]
struct GetClusterStatusResponse {
node: String,
garage_version: &'static str,
garage_features: Option<&'static [&'static str]>,
db_engine: String,
known_nodes: HashMap<String, KnownNodeResp>,
layout: GetClusterLayoutResponse,
}
#[derive(Serialize)]
struct ConnectClusterNodesResponse {
success: bool,
error: Option<String>,
}
#[derive(Serialize)]
#[serde(rename_all = "camelCase")]
struct GetClusterLayoutResponse {
version: u64,
roles: HashMap<String, Option<NodeRole>>,
staged_role_changes: HashMap<String, Option<NodeRole>>,
}
#[derive(Serialize)]
struct KnownNodeResp {
addr: SocketAddr,
is_up: bool,
last_seen_secs_ago: Option<u64>,
hostname: String,
}
pub async fn handle_update_cluster_layout(
garage: &Arc<Garage>,
req: Request<Body>,
) -> Result<Response<Body>, Error> {
let updates = parse_json_body::<UpdateClusterLayoutRequest>(req).await?;
let mut layout = garage.system.get_cluster_layout();
let mut roles = layout.roles.clone();
roles.merge(&layout.staging);
for (node, role) in updates {
let node = hex::decode(node).ok_or_bad_request("Invalid node identifier")?;
let node = Uuid::try_from(&node).ok_or_bad_request("Invalid node identifier")?;
layout
.staging
.merge(&roles.update_mutator(node, NodeRoleV(role)));
}
garage.system.update_cluster_layout(&layout).await?;
Ok(Response::builder()
.status(StatusCode::OK)
.body(Body::empty())?)
}
pub async fn handle_apply_cluster_layout(
garage: &Arc<Garage>,
req: Request<Body>,
) -> Result<Response<Body>, Error> {
let param = parse_json_body::<ApplyRevertLayoutRequest>(req).await?;
let layout = garage.system.get_cluster_layout();
let layout = layout.apply_staged_changes(Some(param.version))?;
garage.system.update_cluster_layout(&layout).await?;
Ok(Response::builder()
.status(StatusCode::OK)
.body(Body::empty())?)
}
pub async fn handle_revert_cluster_layout(
garage: &Arc<Garage>,
req: Request<Body>,
) -> Result<Response<Body>, Error> {
let param = parse_json_body::<ApplyRevertLayoutRequest>(req).await?;
let layout = garage.system.get_cluster_layout();
let layout = layout.revert_staged_changes(Some(param.version))?;
garage.system.update_cluster_layout(&layout).await?;
Ok(Response::builder()
.status(StatusCode::OK)
.body(Body::empty())?)
}
type UpdateClusterLayoutRequest = HashMap<String, Option<NodeRole>>;
#[derive(Deserialize)]
struct ApplyRevertLayoutRequest {
version: u64,
}
+97
View File
@@ -0,0 +1,97 @@
use err_derive::Error;
use hyper::header::HeaderValue;
use hyper::{Body, HeaderMap, StatusCode};
pub use garage_model::helper::error::Error as HelperError;
use crate::common_error::CommonError;
pub use crate::common_error::{CommonErrorDerivative, OkOrBadRequest, OkOrInternalError};
use crate::generic_server::ApiError;
use crate::helpers::CustomApiErrorBody;
/// Errors of this crate
#[derive(Debug, Error)]
pub enum Error {
#[error(display = "{}", _0)]
/// Error from common error
Common(CommonError),
// Category: cannot process
/// The API access key does not exist
#[error(display = "Access key not found: {}", _0)]
NoSuchAccessKey(String),
/// In Import key, the key already exists
#[error(
display = "Key {} already exists in data store. Even if it is deleted, we can't let you create a new key with the same ID. Sorry.",
_0
)]
KeyAlreadyExists(String),
}
impl<T> From<T> for Error
where
CommonError: From<T>,
{
fn from(err: T) -> Self {
Error::Common(CommonError::from(err))
}
}
impl CommonErrorDerivative for Error {}
impl From<HelperError> for Error {
fn from(err: HelperError) -> Self {
match err {
HelperError::Internal(i) => Self::Common(CommonError::InternalError(i)),
HelperError::BadRequest(b) => Self::Common(CommonError::BadRequest(b)),
HelperError::InvalidBucketName(n) => Self::Common(CommonError::InvalidBucketName(n)),
HelperError::NoSuchBucket(n) => Self::Common(CommonError::NoSuchBucket(n)),
HelperError::NoSuchAccessKey(n) => Self::NoSuchAccessKey(n),
}
}
}
impl Error {
fn code(&self) -> &'static str {
match self {
Error::Common(c) => c.aws_code(),
Error::NoSuchAccessKey(_) => "NoSuchAccessKey",
Error::KeyAlreadyExists(_) => "KeyAlreadyExists",
}
}
}
impl ApiError for Error {
/// Get the HTTP status code that best represents the meaning of the error for the client
fn http_status_code(&self) -> StatusCode {
match self {
Error::Common(c) => c.http_status_code(),
Error::NoSuchAccessKey(_) => StatusCode::NOT_FOUND,
Error::KeyAlreadyExists(_) => StatusCode::CONFLICT,
}
}
fn add_http_headers(&self, header_map: &mut HeaderMap<HeaderValue>) {
use hyper::header;
header_map.append(header::CONTENT_TYPE, "application/json".parse().unwrap());
}
fn http_body(&self, garage_region: &str, path: &str) -> Body {
let error = CustomApiErrorBody {
code: self.code().to_string(),
message: format!("{}", self),
path: path.to_string(),
region: garage_region.to_string(),
};
Body::from(serde_json::to_string_pretty(&error).unwrap_or_else(|_| {
r#"
{
"code": "InternalError",
"message": "JSON encoding of error failed"
}
"#
.into()
}))
}
}
+256
View File
@@ -0,0 +1,256 @@
use std::collections::HashMap;
use std::sync::Arc;
use hyper::{Body, Request, Response, StatusCode};
use serde::{Deserialize, Serialize};
use garage_table::*;
use garage_model::garage::Garage;
use garage_model::key_table::*;
use crate::admin::error::*;
use crate::helpers::{json_ok_response, parse_json_body};
pub async fn handle_list_keys(garage: &Arc<Garage>) -> Result<Response<Body>, Error> {
let res = garage
.key_table
.get_range(
&EmptyKey,
None,
Some(KeyFilter::Deleted(DeletedFilter::NotDeleted)),
10000,
EnumerationOrder::Forward,
)
.await?
.iter()
.map(|k| ListKeyResultItem {
id: k.key_id.to_string(),
name: k.params().unwrap().name.get().clone(),
})
.collect::<Vec<_>>();
Ok(json_ok_response(&res)?)
}
#[derive(Serialize)]
struct ListKeyResultItem {
id: String,
name: String,
}
pub async fn handle_get_key_info(
garage: &Arc<Garage>,
id: Option<String>,
search: Option<String>,
) -> Result<Response<Body>, Error> {
let key = if let Some(id) = id {
garage.key_helper().get_existing_key(&id).await?
} else if let Some(search) = search {
garage
.key_helper()
.get_existing_matching_key(&search)
.await?
} else {
unreachable!();
};
key_info_results(garage, key).await
}
pub async fn handle_create_key(
garage: &Arc<Garage>,
req: Request<Body>,
) -> Result<Response<Body>, Error> {
let req = parse_json_body::<CreateKeyRequest>(req).await?;
let key = Key::new(&req.name);
garage.key_table.insert(&key).await?;
key_info_results(garage, key).await
}
#[derive(Deserialize)]
struct CreateKeyRequest {
name: String,
}
pub async fn handle_import_key(
garage: &Arc<Garage>,
req: Request<Body>,
) -> Result<Response<Body>, Error> {
let req = parse_json_body::<ImportKeyRequest>(req).await?;
let prev_key = garage.key_table.get(&EmptyKey, &req.access_key_id).await?;
if prev_key.is_some() {
return Err(Error::KeyAlreadyExists(req.access_key_id.to_string()));
}
let imported_key = Key::import(&req.access_key_id, &req.secret_access_key, &req.name);
garage.key_table.insert(&imported_key).await?;
key_info_results(garage, imported_key).await
}
#[derive(Deserialize)]
#[serde(rename_all = "camelCase")]
struct ImportKeyRequest {
access_key_id: String,
secret_access_key: String,
name: String,
}
pub async fn handle_update_key(
garage: &Arc<Garage>,
id: String,
req: Request<Body>,
) -> Result<Response<Body>, Error> {
let req = parse_json_body::<UpdateKeyRequest>(req).await?;
let mut key = garage.key_helper().get_existing_key(&id).await?;
let key_state = key.state.as_option_mut().unwrap();
if let Some(new_name) = req.name {
key_state.name.update(new_name);
}
if let Some(allow) = req.allow {
if allow.create_bucket {
key_state.allow_create_bucket.update(true);
}
}
if let Some(deny) = req.deny {
if deny.create_bucket {
key_state.allow_create_bucket.update(false);
}
}
garage.key_table.insert(&key).await?;
key_info_results(garage, key).await
}
#[derive(Deserialize)]
struct UpdateKeyRequest {
name: Option<String>,
allow: Option<KeyPerm>,
deny: Option<KeyPerm>,
}
pub async fn handle_delete_key(garage: &Arc<Garage>, id: String) -> Result<Response<Body>, Error> {
let mut key = garage.key_helper().get_existing_key(&id).await?;
key.state.as_option().unwrap();
garage.key_helper().delete_key(&mut key).await?;
Ok(Response::builder()
.status(StatusCode::NO_CONTENT)
.body(Body::empty())?)
}
async fn key_info_results(garage: &Arc<Garage>, key: Key) -> Result<Response<Body>, Error> {
let mut relevant_buckets = HashMap::new();
let key_state = key.state.as_option().unwrap();
for id in key_state
.authorized_buckets
.items()
.iter()
.map(|(id, _)| id)
.chain(
key_state
.local_aliases
.items()
.iter()
.filter_map(|(_, _, v)| v.as_ref()),
) {
if !relevant_buckets.contains_key(id) {
if let Some(b) = garage.bucket_table.get(&EmptyKey, id).await? {
if b.state.as_option().is_some() {
relevant_buckets.insert(*id, b);
}
}
}
}
let res = GetKeyInfoResult {
name: key_state.name.get().clone(),
access_key_id: key.key_id.clone(),
secret_access_key: key_state.secret_key.clone(),
permissions: KeyPerm {
create_bucket: *key_state.allow_create_bucket.get(),
},
buckets: relevant_buckets
.into_iter()
.map(|(_, bucket)| {
let state = bucket.state.as_option().unwrap();
KeyInfoBucketResult {
id: hex::encode(bucket.id),
global_aliases: state
.aliases
.items()
.iter()
.filter(|(_, _, a)| *a)
.map(|(n, _, _)| n.to_string())
.collect::<Vec<_>>(),
local_aliases: state
.local_aliases
.items()
.iter()
.filter(|((k, _), _, a)| *a && *k == key.key_id)
.map(|((_, n), _, _)| n.to_string())
.collect::<Vec<_>>(),
permissions: key_state
.authorized_buckets
.get(&bucket.id)
.map(|p| ApiBucketKeyPerm {
read: p.allow_read,
write: p.allow_write,
owner: p.allow_owner,
})
.unwrap_or_default(),
}
})
.collect::<Vec<_>>(),
};
Ok(json_ok_response(&res)?)
}
#[derive(Serialize)]
#[serde(rename_all = "camelCase")]
struct GetKeyInfoResult {
name: String,
access_key_id: String,
secret_access_key: String,
permissions: KeyPerm,
buckets: Vec<KeyInfoBucketResult>,
}
#[derive(Serialize, Deserialize)]
#[serde(rename_all = "camelCase")]
struct KeyPerm {
#[serde(default)]
create_bucket: bool,
}
#[derive(Serialize)]
#[serde(rename_all = "camelCase")]
struct KeyInfoBucketResult {
id: String,
global_aliases: Vec<String>,
local_aliases: Vec<String>,
permissions: ApiBucketKeyPerm,
}
#[derive(Serialize, Deserialize, Default)]
pub(crate) struct ApiBucketKeyPerm {
#[serde(default)]
pub(crate) read: bool,
#[serde(default)]
pub(crate) write: bool,
#[serde(default)]
pub(crate) owner: bool,
}
+7
View File
@@ -0,0 +1,7 @@
pub mod api_server;
mod error;
mod router;
mod bucket;
mod cluster;
mod key;
+145
View File
@@ -0,0 +1,145 @@
use std::borrow::Cow;
use hyper::{Method, Request};
use crate::admin::error::*;
use crate::router_macros::*;
pub enum Authorization {
MetricsToken,
AdminToken,
}
router_match! {@func
/// List of all Admin API endpoints.
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Endpoint {
Options,
Metrics,
GetClusterStatus,
ConnectClusterNodes,
// Layout
GetClusterLayout,
UpdateClusterLayout,
ApplyClusterLayout,
RevertClusterLayout,
// Keys
ListKeys,
CreateKey,
ImportKey,
GetKeyInfo {
id: Option<String>,
search: Option<String>,
},
DeleteKey {
id: String,
},
UpdateKey {
id: String,
},
// Buckets
ListBuckets,
CreateBucket,
GetBucketInfo {
id: Option<String>,
global_alias: Option<String>,
},
DeleteBucket {
id: String,
},
UpdateBucket {
id: String,
},
// Bucket-Key Permissions
BucketAllowKey,
BucketDenyKey,
// Bucket aliases
GlobalAliasBucket {
id: String,
alias: String,
},
GlobalUnaliasBucket {
id: String,
alias: String,
},
LocalAliasBucket {
id: String,
access_key_id: String,
alias: String,
},
LocalUnaliasBucket {
id: String,
access_key_id: String,
alias: String,
},
}}
impl Endpoint {
/// Determine which S3 endpoint a request is for using the request, and a bucket which was
/// possibly extracted from the Host header.
/// Returns Self plus bucket name, if endpoint is not Endpoint::ListBuckets
pub fn from_request<T>(req: &Request<T>) -> Result<Self, Error> {
let uri = req.uri();
let path = uri.path();
let query = uri.query();
let mut query = QueryParameters::from_query(query.unwrap_or_default())?;
let res = router_match!(@gen_path_parser (req.method(), path, query) [
OPTIONS _ => Options,
GET "/metrics" => Metrics,
GET "/v0/status" => GetClusterStatus,
POST "/v0/connect" => ConnectClusterNodes,
// Layout endpoints
GET "/v0/layout" => GetClusterLayout,
POST "/v0/layout" => UpdateClusterLayout,
POST "/v0/layout/apply" => ApplyClusterLayout,
POST "/v0/layout/revert" => RevertClusterLayout,
// API key endpoints
GET "/v0/key" if id => GetKeyInfo (query_opt::id, query_opt::search),
GET "/v0/key" if search => GetKeyInfo (query_opt::id, query_opt::search),
POST "/v0/key" if id => UpdateKey (query::id),
POST "/v0/key" => CreateKey,
POST "/v0/key/import" => ImportKey,
DELETE "/v0/key" if id => DeleteKey (query::id),
GET "/v0/key" => ListKeys,
// Bucket endpoints
GET "/v0/bucket" if id => GetBucketInfo (query_opt::id, query_opt::global_alias),
GET "/v0/bucket" if global_alias => GetBucketInfo (query_opt::id, query_opt::global_alias),
GET "/v0/bucket" => ListBuckets,
POST "/v0/bucket" => CreateBucket,
DELETE "/v0/bucket" if id => DeleteBucket (query::id),
PUT "/v0/bucket" if id => UpdateBucket (query::id),
// Bucket-key permissions
POST "/v0/bucket/allow" => BucketAllowKey,
POST "/v0/bucket/deny" => BucketDenyKey,
// Bucket aliases
PUT "/v0/bucket/alias/global" => GlobalAliasBucket (query::id, query::alias),
DELETE "/v0/bucket/alias/global" => GlobalUnaliasBucket (query::id, query::alias),
PUT "/v0/bucket/alias/local" => LocalAliasBucket (query::id, query::access_key_id, query::alias),
DELETE "/v0/bucket/alias/local" => LocalUnaliasBucket (query::id, query::access_key_id, query::alias),
]);
if let Some(message) = query.nonempty_message() {
debug!("Unused query parameter: {}", message)
}
Ok(res)
}
/// Get the kind of authorization which is required to perform the operation.
pub fn authorization_type(&self) -> Authorization {
match self {
Self::Metrics => Authorization::MetricsToken,
_ => Authorization::AdminToken,
}
}
}
generateQueryParameters! {
"id" => id,
"search" => search,
"globalAlias" => global_alias,
"alias" => alias,
"accessKeyId" => access_key_id
}
-645
View File
@@ -1,645 +0,0 @@
use std::net::SocketAddr;
use std::sync::Arc;
use chrono::{DateTime, NaiveDateTime, Utc};
use futures::future::Future;
use futures::prelude::*;
use hyper::header;
use hyper::server::conn::AddrStream;
use hyper::service::{make_service_fn, service_fn};
use hyper::{Body, Method, Request, Response, Server};
use opentelemetry::{
global,
metrics::{Counter, ValueRecorder},
trace::{FutureExt, TraceContextExt, Tracer},
Context, KeyValue,
};
use garage_util::data::*;
use garage_util::error::Error as GarageError;
use garage_util::metrics::{gen_trace_id, RecordDuration};
use garage_model::garage::Garage;
use garage_model::key_table::Key;
use garage_table::util::*;
use crate::error::*;
use crate::signature::compute_scope;
use crate::signature::payload::check_payload_signature;
use crate::signature::streaming::SignedPayloadStream;
use crate::signature::LONG_DATETIME;
use crate::helpers::*;
use crate::s3_bucket::*;
use crate::s3_copy::*;
use crate::s3_cors::*;
use crate::s3_delete::*;
use crate::s3_get::*;
use crate::s3_list::*;
use crate::s3_post_object::handle_post_object;
use crate::s3_put::*;
use crate::s3_router::{Authorization, Endpoint};
use crate::s3_website::*;
struct ApiMetrics {
request_counter: Counter<u64>,
error_counter: Counter<u64>,
request_duration: ValueRecorder<f64>,
}
impl ApiMetrics {
fn new() -> Self {
let meter = global::meter("garage/api");
Self {
request_counter: meter
.u64_counter("api.request_counter")
.with_description("Number of API calls to the various S3 API endpoints")
.init(),
error_counter: meter
.u64_counter("api.error_counter")
.with_description(
"Number of API calls to the various S3 API endpoints that resulted in errors",
)
.init(),
request_duration: meter
.f64_value_recorder("api.request_duration")
.with_description("Duration of API calls to the various S3 API endpoints")
.init(),
}
}
}
/// Run the S3 API server
pub async fn run_api_server(
garage: Arc<Garage>,
shutdown_signal: impl Future<Output = ()>,
) -> Result<(), GarageError> {
let addr = &garage.config.s3_api.api_bind_addr;
let metrics = Arc::new(ApiMetrics::new());
let service = make_service_fn(|conn: &AddrStream| {
let garage = garage.clone();
let metrics = metrics.clone();
let client_addr = conn.remote_addr();
async move {
Ok::<_, GarageError>(service_fn(move |req: Request<Body>| {
let garage = garage.clone();
let metrics = metrics.clone();
handler(garage, metrics, req, client_addr)
}))
}
});
let server = Server::bind(addr).serve(service);
let graceful = server.with_graceful_shutdown(shutdown_signal);
info!("API server listening on http://{}", addr);
graceful.await?;
Ok(())
}
async fn handler(
garage: Arc<Garage>,
metrics: Arc<ApiMetrics>,
req: Request<Body>,
addr: SocketAddr,
) -> Result<Response<Body>, GarageError> {
let uri = req.uri().clone();
info!("{} {} {}", addr, req.method(), uri);
debug!("{:?}", req);
let tracer = opentelemetry::global::tracer("garage");
let span = tracer
.span_builder("S3 API call (unknown)")
.with_trace_id(gen_trace_id())
.with_attributes(vec![
KeyValue::new("method", format!("{}", req.method())),
KeyValue::new("uri", req.uri().to_string()),
])
.start(&tracer);
let res = handler_stage2(garage.clone(), metrics, req)
.with_context(Context::current_with_span(span))
.await;
match res {
Ok(x) => {
debug!("{} {:?}", x.status(), x.headers());
Ok(x)
}
Err(e) => {
let body: Body = Body::from(e.aws_xml(&garage.config.s3_api.s3_region, uri.path()));
let mut http_error_builder = Response::builder()
.status(e.http_status_code())
.header("Content-Type", "application/xml");
if let Some(header_map) = http_error_builder.headers_mut() {
e.add_headers(header_map)
}
let http_error = http_error_builder.body(body)?;
if e.http_status_code().is_server_error() {
warn!("Response: error {}, {}", e.http_status_code(), e);
} else {
info!("Response: error {}, {}", e.http_status_code(), e);
}
Ok(http_error)
}
}
}
async fn handler_stage2(
garage: Arc<Garage>,
metrics: Arc<ApiMetrics>,
req: Request<Body>,
) -> Result<Response<Body>, Error> {
let authority = req
.headers()
.get(header::HOST)
.ok_or_bad_request("Host header required")?
.to_str()?;
let host = authority_to_host(authority)?;
let bucket_name = garage
.config
.s3_api
.root_domain
.as_ref()
.and_then(|root_domain| host_to_bucket(&host, root_domain));
let (endpoint, bucket_name) = Endpoint::from_request(&req, bucket_name.map(ToOwned::to_owned))?;
debug!("Endpoint: {:?}", endpoint);
let current_context = Context::current();
let current_span = current_context.span();
current_span.update_name::<String>(format!("S3 API {}", endpoint.name()));
current_span.set_attribute(KeyValue::new("endpoint", endpoint.name()));
current_span.set_attribute(KeyValue::new(
"bucket",
bucket_name.clone().unwrap_or_default(),
));
let metrics_tags = &[KeyValue::new("api_endpoint", endpoint.name())];
let res = handler_stage3(garage, req, endpoint, bucket_name)
.record_duration(&metrics.request_duration, &metrics_tags[..])
.await;
metrics.request_counter.add(1, &metrics_tags[..]);
let status_code = match &res {
Ok(r) => r.status(),
Err(e) => e.http_status_code(),
};
if status_code.is_client_error() || status_code.is_server_error() {
metrics.error_counter.add(
1,
&[
metrics_tags[0].clone(),
KeyValue::new("status_code", status_code.as_str().to_string()),
],
);
}
res
}
async fn handler_stage3(
garage: Arc<Garage>,
req: Request<Body>,
endpoint: Endpoint,
bucket_name: Option<String>,
) -> Result<Response<Body>, Error> {
// Some endpoints are processed early, before we even check for an API key
if let Endpoint::PostObject = endpoint {
return handle_post_object(garage, req, bucket_name.unwrap()).await;
}
if let Endpoint::Options = endpoint {
return handle_options_s3api(garage, &req, bucket_name).await;
}
let (api_key, mut content_sha256) = check_payload_signature(&garage, &req).await?;
let api_key = api_key.ok_or_else(|| {
Error::Forbidden("Garage does not support anonymous access yet".to_string())
})?;
let req = match req.headers().get("x-amz-content-sha256") {
Some(header) if header == "STREAMING-AWS4-HMAC-SHA256-PAYLOAD" => {
let signature = content_sha256
.take()
.ok_or_bad_request("No signature provided")?;
let secret_key = &api_key
.state
.as_option()
.ok_or_internal_error("Deleted key state")?
.secret_key;
let date = req
.headers()
.get("x-amz-date")
.ok_or_bad_request("Missing X-Amz-Date field")?
.to_str()?;
let date: NaiveDateTime = NaiveDateTime::parse_from_str(date, LONG_DATETIME)
.ok_or_bad_request("Invalid date")?;
let date: DateTime<Utc> = DateTime::from_utc(date, Utc);
let scope = compute_scope(&date, &garage.config.s3_api.s3_region);
let signing_hmac = crate::signature::signing_hmac(
&date,
secret_key,
&garage.config.s3_api.s3_region,
"s3",
)
.ok_or_internal_error("Unable to build signing HMAC")?;
req.map(move |body| {
Body::wrap_stream(
SignedPayloadStream::new(
body.map_err(Error::from),
signing_hmac,
date,
&scope,
signature,
)
.map_err(Error::from),
)
})
}
_ => req,
};
let bucket_name = match bucket_name {
None => return handle_request_without_bucket(garage, req, api_key, endpoint).await,
Some(bucket) => bucket.to_string(),
};
// Special code path for CreateBucket API endpoint
if let Endpoint::CreateBucket {} = endpoint {
return handle_create_bucket(&garage, req, content_sha256, api_key, bucket_name).await;
}
let bucket_id = resolve_bucket(&garage, &bucket_name, &api_key).await?;
let bucket = garage
.bucket_table
.get(&EmptyKey, &bucket_id)
.await?
.filter(|b| !b.state.is_deleted())
.ok_or(Error::NoSuchBucket)?;
let allowed = match endpoint.authorization_type() {
Authorization::Read => api_key.allow_read(&bucket_id),
Authorization::Write => api_key.allow_write(&bucket_id),
Authorization::Owner => api_key.allow_owner(&bucket_id),
_ => unreachable!(),
};
if !allowed {
return Err(Error::Forbidden(
"Operation is not allowed for this key.".to_string(),
));
}
// Look up what CORS rule might apply to response.
// Requests for methods different than GET, HEAD or POST
// are always preflighted, i.e. the browser should make
// an OPTIONS call before to check it is allowed
let matching_cors_rule = match *req.method() {
Method::GET | Method::HEAD | Method::POST => find_matching_cors_rule(&bucket, &req)?,
_ => None,
};
let resp = match endpoint {
Endpoint::HeadObject {
key, part_number, ..
} => handle_head(garage, &req, bucket_id, &key, part_number).await,
Endpoint::GetObject {
key, part_number, ..
} => handle_get(garage, &req, bucket_id, &key, part_number).await,
Endpoint::UploadPart {
key,
part_number,
upload_id,
} => {
handle_put_part(
garage,
req,
bucket_id,
&key,
part_number,
&upload_id,
content_sha256,
)
.await
}
Endpoint::CopyObject { key } => handle_copy(garage, &api_key, &req, bucket_id, &key).await,
Endpoint::UploadPartCopy {
key,
part_number,
upload_id,
} => {
handle_upload_part_copy(
garage,
&api_key,
&req,
bucket_id,
&key,
part_number,
&upload_id,
)
.await
}
Endpoint::PutObject { key } => {
handle_put(garage, req, bucket_id, &key, content_sha256).await
}
Endpoint::AbortMultipartUpload { key, upload_id } => {
handle_abort_multipart_upload(garage, bucket_id, &key, &upload_id).await
}
Endpoint::DeleteObject { key, .. } => handle_delete(garage, bucket_id, &key).await,
Endpoint::CreateMultipartUpload { key } => {
handle_create_multipart_upload(garage, &req, &bucket_name, bucket_id, &key).await
}
Endpoint::CompleteMultipartUpload { key, upload_id } => {
handle_complete_multipart_upload(
garage,
req,
&bucket_name,
bucket_id,
&key,
&upload_id,
content_sha256,
)
.await
}
Endpoint::CreateBucket {} => unreachable!(),
Endpoint::HeadBucket {} => {
let empty_body: Body = Body::from(vec![]);
let response = Response::builder().body(empty_body).unwrap();
Ok(response)
}
Endpoint::DeleteBucket {} => {
handle_delete_bucket(&garage, bucket_id, bucket_name, api_key).await
}
Endpoint::GetBucketLocation {} => handle_get_bucket_location(garage),
Endpoint::GetBucketVersioning {} => handle_get_bucket_versioning(),
Endpoint::ListObjects {
delimiter,
encoding_type,
marker,
max_keys,
prefix,
} => {
handle_list(
garage,
&ListObjectsQuery {
common: ListQueryCommon {
bucket_name,
bucket_id,
delimiter: delimiter.map(|d| d.to_string()),
page_size: max_keys.map(|p| p.clamp(1, 1000)).unwrap_or(1000),
prefix: prefix.unwrap_or_default(),
urlencode_resp: encoding_type.map(|e| e == "url").unwrap_or(false),
},
is_v2: false,
marker,
continuation_token: None,
start_after: None,
},
)
.await
}
Endpoint::ListObjectsV2 {
delimiter,
encoding_type,
max_keys,
prefix,
continuation_token,
start_after,
list_type,
..
} => {
if list_type == "2" {
handle_list(
garage,
&ListObjectsQuery {
common: ListQueryCommon {
bucket_name,
bucket_id,
delimiter: delimiter.map(|d| d.to_string()),
page_size: max_keys.map(|p| p.clamp(1, 1000)).unwrap_or(1000),
urlencode_resp: encoding_type.map(|e| e == "url").unwrap_or(false),
prefix: prefix.unwrap_or_default(),
},
is_v2: true,
marker: None,
continuation_token,
start_after,
},
)
.await
} else {
Err(Error::BadRequest(format!(
"Invalid endpoint: list-type={}",
list_type
)))
}
}
Endpoint::ListMultipartUploads {
delimiter,
encoding_type,
key_marker,
max_uploads,
prefix,
upload_id_marker,
} => {
handle_list_multipart_upload(
garage,
&ListMultipartUploadsQuery {
common: ListQueryCommon {
bucket_name,
bucket_id,
delimiter: delimiter.map(|d| d.to_string()),
page_size: max_uploads.map(|p| p.clamp(1, 1000)).unwrap_or(1000),
prefix: prefix.unwrap_or_default(),
urlencode_resp: encoding_type.map(|e| e == "url").unwrap_or(false),
},
key_marker,
upload_id_marker,
},
)
.await
}
Endpoint::ListParts {
key,
max_parts,
part_number_marker,
upload_id,
} => {
handle_list_parts(
garage,
&ListPartsQuery {
bucket_name,
bucket_id,
key,
upload_id,
part_number_marker: part_number_marker.map(|p| p.clamp(1, 10000)),
max_parts: max_parts.map(|p| p.clamp(1, 1000)).unwrap_or(1000),
},
)
.await
}
Endpoint::DeleteObjects {} => {
handle_delete_objects(garage, bucket_id, req, content_sha256).await
}
Endpoint::GetBucketWebsite {} => handle_get_website(&bucket).await,
Endpoint::PutBucketWebsite {} => {
handle_put_website(garage, bucket_id, req, content_sha256).await
}
Endpoint::DeleteBucketWebsite {} => handle_delete_website(garage, bucket_id).await,
Endpoint::GetBucketCors {} => handle_get_cors(&bucket).await,
Endpoint::PutBucketCors {} => handle_put_cors(garage, bucket_id, req, content_sha256).await,
Endpoint::DeleteBucketCors {} => handle_delete_cors(garage, bucket_id).await,
endpoint => Err(Error::NotImplemented(endpoint.name().to_owned())),
};
// If request was a success and we have a CORS rule that applies to it,
// add the corresponding CORS headers to the response
let mut resp_ok = resp?;
if let Some(rule) = matching_cors_rule {
add_cors_headers(&mut resp_ok, rule)
.ok_or_internal_error("Invalid bucket CORS configuration")?;
}
Ok(resp_ok)
}
async fn handle_request_without_bucket(
garage: Arc<Garage>,
_req: Request<Body>,
api_key: Key,
endpoint: Endpoint,
) -> Result<Response<Body>, Error> {
match endpoint {
Endpoint::ListBuckets => handle_list_buckets(&garage, &api_key).await,
endpoint => Err(Error::NotImplemented(endpoint.name().to_owned())),
}
}
#[allow(clippy::ptr_arg)]
pub async fn resolve_bucket(
garage: &Garage,
bucket_name: &String,
api_key: &Key,
) -> Result<Uuid, Error> {
let api_key_params = api_key
.state
.as_option()
.ok_or_internal_error("Key should not be deleted at this point")?;
if let Some(Some(bucket_id)) = api_key_params.local_aliases.get(bucket_name) {
Ok(*bucket_id)
} else {
Ok(garage
.bucket_helper()
.resolve_global_bucket_name(bucket_name)
.await?
.ok_or(Error::NoSuchBucket)?)
}
}
/// Extract the bucket name and the key name from an HTTP path and possibly a bucket provided in
/// the host header of the request
///
/// S3 internally manages only buckets and keys. This function splits
/// an HTTP path to get the corresponding bucket name and key.
pub fn parse_bucket_key<'a>(
path: &'a str,
host_bucket: Option<&'a str>,
) -> Result<(&'a str, Option<&'a str>), Error> {
let path = path.trim_start_matches('/');
if let Some(bucket) = host_bucket {
if !path.is_empty() {
return Ok((bucket, Some(path)));
} else {
return Ok((bucket, None));
}
}
let (bucket, key) = match path.find('/') {
Some(i) => {
let key = &path[i + 1..];
if !key.is_empty() {
(&path[..i], Some(key))
} else {
(&path[..i], None)
}
}
None => (path, None),
};
if bucket.is_empty() {
return Err(Error::BadRequest("No bucket specified".to_string()));
}
Ok((bucket, key))
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn parse_bucket_containing_a_key() -> Result<(), Error> {
let (bucket, key) = parse_bucket_key("/my_bucket/a/super/file.jpg", None)?;
assert_eq!(bucket, "my_bucket");
assert_eq!(key.expect("key must be set"), "a/super/file.jpg");
Ok(())
}
#[test]
fn parse_bucket_containing_no_key() -> Result<(), Error> {
let (bucket, key) = parse_bucket_key("/my_bucket/", None)?;
assert_eq!(bucket, "my_bucket");
assert!(key.is_none());
let (bucket, key) = parse_bucket_key("/my_bucket", None)?;
assert_eq!(bucket, "my_bucket");
assert!(key.is_none());
Ok(())
}
#[test]
fn parse_bucket_containing_no_bucket() {
let parsed = parse_bucket_key("", None);
assert!(parsed.is_err());
let parsed = parse_bucket_key("/", None);
assert!(parsed.is_err());
let parsed = parse_bucket_key("////", None);
assert!(parsed.is_err());
}
#[test]
fn parse_bucket_with_vhost_and_key() -> Result<(), Error> {
let (bucket, key) = parse_bucket_key("/a/super/file.jpg", Some("my-bucket"))?;
assert_eq!(bucket, "my-bucket");
assert_eq!(key.expect("key must be set"), "a/super/file.jpg");
Ok(())
}
#[test]
fn parse_bucket_with_vhost_no_key() -> Result<(), Error> {
let (bucket, key) = parse_bucket_key("", Some("my-bucket"))?;
assert_eq!(bucket, "my-bucket");
assert!(key.is_none());
let (bucket, key) = parse_bucket_key("/", Some("my-bucket"))?;
assert_eq!(bucket, "my-bucket");
assert!(key.is_none());
Ok(())
}
}
+177
View File
@@ -0,0 +1,177 @@
use err_derive::Error;
use hyper::StatusCode;
use garage_util::error::Error as GarageError;
/// Errors of this crate
#[derive(Debug, Error)]
pub enum CommonError {
// ---- INTERNAL ERRORS ----
/// Error related to deeper parts of Garage
#[error(display = "Internal error: {}", _0)]
InternalError(#[error(source)] GarageError),
/// Error related to Hyper
#[error(display = "Internal error (Hyper error): {}", _0)]
Hyper(#[error(source)] hyper::Error),
/// Error related to HTTP
#[error(display = "Internal error (HTTP error): {}", _0)]
Http(#[error(source)] http::Error),
// ---- GENERIC CLIENT ERRORS ----
/// Proper authentication was not provided
#[error(display = "Forbidden: {}", _0)]
Forbidden(String),
/// Generic bad request response with custom message
#[error(display = "Bad request: {}", _0)]
BadRequest(String),
// ---- SPECIFIC ERROR CONDITIONS ----
// These have to be error codes referenced in the S3 spec here:
// https://docs.aws.amazon.com/AmazonS3/latest/API/ErrorResponses.html#ErrorCodeList
/// The bucket requested don't exists
#[error(display = "Bucket not found: {}", _0)]
NoSuchBucket(String),
/// Tried to create a bucket that already exist
#[error(display = "Bucket already exists")]
BucketAlreadyExists,
/// Tried to delete a non-empty bucket
#[error(display = "Tried to delete a non-empty bucket")]
BucketNotEmpty,
// Category: bad request
/// Bucket name is not valid according to AWS S3 specs
#[error(display = "Invalid bucket name: {}", _0)]
InvalidBucketName(String),
}
impl CommonError {
pub fn http_status_code(&self) -> StatusCode {
match self {
CommonError::InternalError(
GarageError::Timeout
| GarageError::RemoteError(_)
| GarageError::Quorum(_, _, _, _),
) => StatusCode::SERVICE_UNAVAILABLE,
CommonError::InternalError(_) | CommonError::Hyper(_) | CommonError::Http(_) => {
StatusCode::INTERNAL_SERVER_ERROR
}
CommonError::BadRequest(_) => StatusCode::BAD_REQUEST,
CommonError::Forbidden(_) => StatusCode::FORBIDDEN,
CommonError::NoSuchBucket(_) => StatusCode::NOT_FOUND,
CommonError::BucketNotEmpty | CommonError::BucketAlreadyExists => StatusCode::CONFLICT,
CommonError::InvalidBucketName(_) => StatusCode::BAD_REQUEST,
}
}
pub fn aws_code(&self) -> &'static str {
match self {
CommonError::Forbidden(_) => "AccessDenied",
CommonError::InternalError(
GarageError::Timeout
| GarageError::RemoteError(_)
| GarageError::Quorum(_, _, _, _),
) => "ServiceUnavailable",
CommonError::InternalError(_) | CommonError::Hyper(_) | CommonError::Http(_) => {
"InternalError"
}
CommonError::BadRequest(_) => "InvalidRequest",
CommonError::NoSuchBucket(_) => "NoSuchBucket",
CommonError::BucketAlreadyExists => "BucketAlreadyExists",
CommonError::BucketNotEmpty => "BucketNotEmpty",
CommonError::InvalidBucketName(_) => "InvalidBucketName",
}
}
pub fn bad_request<M: ToString>(msg: M) -> Self {
CommonError::BadRequest(msg.to_string())
}
}
pub trait CommonErrorDerivative: From<CommonError> {
fn internal_error<M: ToString>(msg: M) -> Self {
Self::from(CommonError::InternalError(GarageError::Message(
msg.to_string(),
)))
}
fn bad_request<M: ToString>(msg: M) -> Self {
Self::from(CommonError::BadRequest(msg.to_string()))
}
fn forbidden<M: ToString>(msg: M) -> Self {
Self::from(CommonError::Forbidden(msg.to_string()))
}
}
/// Trait to map error to the Bad Request error code
pub trait OkOrBadRequest {
type S;
fn ok_or_bad_request<M: AsRef<str>>(self, reason: M) -> Result<Self::S, CommonError>;
}
impl<T, E> OkOrBadRequest for Result<T, E>
where
E: std::fmt::Display,
{
type S = T;
fn ok_or_bad_request<M: AsRef<str>>(self, reason: M) -> Result<T, CommonError> {
match self {
Ok(x) => Ok(x),
Err(e) => Err(CommonError::BadRequest(format!(
"{}: {}",
reason.as_ref(),
e
))),
}
}
}
impl<T> OkOrBadRequest for Option<T> {
type S = T;
fn ok_or_bad_request<M: AsRef<str>>(self, reason: M) -> Result<T, CommonError> {
match self {
Some(x) => Ok(x),
None => Err(CommonError::BadRequest(reason.as_ref().to_string())),
}
}
}
/// Trait to map an error to an Internal Error code
pub trait OkOrInternalError {
type S;
fn ok_or_internal_error<M: AsRef<str>>(self, reason: M) -> Result<Self::S, CommonError>;
}
impl<T, E> OkOrInternalError for Result<T, E>
where
E: std::fmt::Display,
{
type S = T;
fn ok_or_internal_error<M: AsRef<str>>(self, reason: M) -> Result<T, CommonError> {
match self {
Ok(x) => Ok(x),
Err(e) => Err(CommonError::InternalError(GarageError::Message(format!(
"{}: {}",
reason.as_ref(),
e
)))),
}
}
}
impl<T> OkOrInternalError for Option<T> {
type S = T;
fn ok_or_internal_error<M: AsRef<str>>(self, reason: M) -> Result<T, CommonError> {
match self {
Some(x) => Ok(x),
None => Err(CommonError::InternalError(GarageError::Message(
reason.as_ref().to_string(),
))),
}
}
}
+211
View File
@@ -0,0 +1,211 @@
use std::net::SocketAddr;
use std::sync::Arc;
use async_trait::async_trait;
use futures::future::Future;
use hyper::header::HeaderValue;
use hyper::server::conn::AddrStream;
use hyper::service::{make_service_fn, service_fn};
use hyper::{Body, Request, Response, Server};
use hyper::{HeaderMap, StatusCode};
use opentelemetry::{
global,
metrics::{Counter, ValueRecorder},
trace::{FutureExt, SpanRef, TraceContextExt, Tracer},
Context, KeyValue,
};
use garage_util::error::Error as GarageError;
use garage_util::metrics::{gen_trace_id, RecordDuration};
pub(crate) trait ApiEndpoint: Send + Sync + 'static {
fn name(&self) -> &'static str;
fn add_span_attributes(&self, span: SpanRef<'_>);
}
pub trait ApiError: std::error::Error + Send + Sync + 'static {
fn http_status_code(&self) -> StatusCode;
fn add_http_headers(&self, header_map: &mut HeaderMap<HeaderValue>);
fn http_body(&self, garage_region: &str, path: &str) -> Body;
}
#[async_trait]
pub(crate) trait ApiHandler: Send + Sync + 'static {
const API_NAME: &'static str;
const API_NAME_DISPLAY: &'static str;
type Endpoint: ApiEndpoint;
type Error: ApiError;
fn parse_endpoint(&self, r: &Request<Body>) -> Result<Self::Endpoint, Self::Error>;
async fn handle(
&self,
req: Request<Body>,
endpoint: Self::Endpoint,
) -> Result<Response<Body>, Self::Error>;
}
pub(crate) struct ApiServer<A: ApiHandler> {
region: String,
api_handler: A,
// Metrics
request_counter: Counter<u64>,
error_counter: Counter<u64>,
request_duration: ValueRecorder<f64>,
}
impl<A: ApiHandler> ApiServer<A> {
pub fn new(region: String, api_handler: A) -> Arc<Self> {
let meter = global::meter("garage/api");
Arc::new(Self {
region,
api_handler,
request_counter: meter
.u64_counter(format!("api.{}.request_counter", A::API_NAME))
.with_description(format!(
"Number of API calls to the various {} API endpoints",
A::API_NAME_DISPLAY
))
.init(),
error_counter: meter
.u64_counter(format!("api.{}.error_counter", A::API_NAME))
.with_description(format!(
"Number of API calls to the various {} API endpoints that resulted in errors",
A::API_NAME_DISPLAY
))
.init(),
request_duration: meter
.f64_value_recorder(format!("api.{}.request_duration", A::API_NAME))
.with_description(format!(
"Duration of API calls to the various {} API endpoints",
A::API_NAME_DISPLAY
))
.init(),
})
}
pub async fn run_server(
self: Arc<Self>,
bind_addr: SocketAddr,
shutdown_signal: impl Future<Output = ()>,
) -> Result<(), GarageError> {
let service = make_service_fn(|conn: &AddrStream| {
let this = self.clone();
let client_addr = conn.remote_addr();
async move {
Ok::<_, GarageError>(service_fn(move |req: Request<Body>| {
let this = this.clone();
this.handler(req, client_addr)
}))
}
});
let server = Server::bind(&bind_addr).serve(service);
let graceful = server.with_graceful_shutdown(shutdown_signal);
info!(
"{} API server listening on http://{}",
A::API_NAME_DISPLAY,
bind_addr
);
graceful.await?;
Ok(())
}
async fn handler(
self: Arc<Self>,
req: Request<Body>,
addr: SocketAddr,
) -> Result<Response<Body>, GarageError> {
let uri = req.uri().clone();
info!("{} {} {}", addr, req.method(), uri);
debug!("{:?}", req);
let tracer = opentelemetry::global::tracer("garage");
let span = tracer
.span_builder(format!("{} API call (unknown)", A::API_NAME_DISPLAY))
.with_trace_id(gen_trace_id())
.with_attributes(vec![
KeyValue::new("method", format!("{}", req.method())),
KeyValue::new("uri", req.uri().to_string()),
])
.start(&tracer);
let res = self
.handler_stage2(req)
.with_context(Context::current_with_span(span))
.await;
match res {
Ok(x) => {
debug!("{} {:?}", x.status(), x.headers());
Ok(x)
}
Err(e) => {
let body: Body = e.http_body(&self.region, uri.path());
let mut http_error_builder = Response::builder().status(e.http_status_code());
if let Some(header_map) = http_error_builder.headers_mut() {
e.add_http_headers(header_map)
}
let http_error = http_error_builder.body(body)?;
if e.http_status_code().is_server_error() {
warn!("Response: error {}, {}", e.http_status_code(), e);
} else {
info!("Response: error {}, {}", e.http_status_code(), e);
}
Ok(http_error)
}
}
}
async fn handler_stage2(&self, req: Request<Body>) -> Result<Response<Body>, A::Error> {
let endpoint = self.api_handler.parse_endpoint(&req)?;
debug!("Endpoint: {}", endpoint.name());
let current_context = Context::current();
let current_span = current_context.span();
current_span.update_name::<String>(format!(
"{} API {}",
A::API_NAME_DISPLAY,
endpoint.name()
));
current_span.set_attribute(KeyValue::new("endpoint", endpoint.name()));
endpoint.add_span_attributes(current_span);
let metrics_tags = &[KeyValue::new("api_endpoint", endpoint.name())];
let res = self
.api_handler
.handle(req, endpoint)
.record_duration(&self.request_duration, &metrics_tags[..])
.await;
self.request_counter.add(1, &metrics_tags[..]);
let status_code = match &res {
Ok(r) => r.status(),
Err(e) => e.http_status_code(),
};
if status_code.is_client_error() || status_code.is_server_error() {
self.error_counter.add(
1,
&[
metrics_tags[0].clone(),
KeyValue::new("status_code", status_code.as_str().to_string()),
],
);
}
res
}
}
+187 -4
View File
@@ -1,5 +1,21 @@
use crate::Error;
use hyper::{Body, Request, Response};
use idna::domain_to_unicode;
use serde::{Deserialize, Serialize};
use crate::common_error::{CommonError as Error, *};
/// What kind of authorization is required to perform a given action
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Authorization {
/// No authorization is required
None,
/// Having Read permission on bucket
Read,
/// Having Write permission on bucket
Write,
/// Having Owner permission on bucket
Owner,
}
/// Host to bucket
///
@@ -31,7 +47,7 @@ pub fn authority_to_host(authority: &str) -> Result<String, Error> {
let mut iter = authority.chars().enumerate();
let (_, first_char) = iter
.next()
.ok_or_else(|| Error::BadRequest("Authority is empty".to_string()))?;
.ok_or_else(|| Error::bad_request("Authority is empty".to_string()))?;
let split = match first_char {
'[' => {
@@ -39,7 +55,7 @@ pub fn authority_to_host(authority: &str) -> Result<String, Error> {
match iter.next() {
Some((_, ']')) => iter.next(),
_ => {
return Err(Error::BadRequest(format!(
return Err(Error::bad_request(format!(
"Authority {} has an illegal format",
authority
)))
@@ -52,7 +68,7 @@ pub fn authority_to_host(authority: &str) -> Result<String, Error> {
let authority = match split {
Some((i, ':')) => Ok(&authority[..i]),
None => Ok(authority),
Some((_, _)) => Err(Error::BadRequest(format!(
Some((_, _)) => Err(Error::bad_request(format!(
"Authority {} has an illegal format",
authority
))),
@@ -60,10 +76,134 @@ pub fn authority_to_host(authority: &str) -> Result<String, Error> {
authority.map(|h| domain_to_unicode(h).0)
}
/// Extract the bucket name and the key name from an HTTP path and possibly a bucket provided in
/// the host header of the request
///
/// S3 internally manages only buckets and keys. This function splits
/// an HTTP path to get the corresponding bucket name and key.
pub fn parse_bucket_key<'a>(
path: &'a str,
host_bucket: Option<&'a str>,
) -> Result<(&'a str, Option<&'a str>), Error> {
let path = path.trim_start_matches('/');
if let Some(bucket) = host_bucket {
if !path.is_empty() {
return Ok((bucket, Some(path)));
} else {
return Ok((bucket, None));
}
}
let (bucket, key) = match path.find('/') {
Some(i) => {
let key = &path[i + 1..];
if !key.is_empty() {
(&path[..i], Some(key))
} else {
(&path[..i], None)
}
}
None => (path, None),
};
if bucket.is_empty() {
return Err(Error::bad_request("No bucket specified"));
}
Ok((bucket, key))
}
const UTF8_BEFORE_LAST_CHAR: char = '\u{10FFFE}';
/// Compute the key after the prefix
pub fn key_after_prefix(pfx: &str) -> Option<String> {
let mut next = pfx.to_string();
while !next.is_empty() {
let tail = next.pop().unwrap();
if tail >= char::MAX {
continue;
}
// Circumvent a limitation of RangeFrom that overflow earlier than needed
// See: https://doc.rust-lang.org/core/ops/struct.RangeFrom.html
let new_tail = if tail == UTF8_BEFORE_LAST_CHAR {
char::MAX
} else {
(tail..).nth(1).unwrap()
};
next.push(new_tail);
return Some(next);
}
None
}
pub async fn parse_json_body<T: for<'de> Deserialize<'de>>(req: Request<Body>) -> Result<T, Error> {
let body = hyper::body::to_bytes(req.into_body()).await?;
let resp: T = serde_json::from_slice(&body).ok_or_bad_request("Invalid JSON")?;
Ok(resp)
}
pub fn json_ok_response<T: Serialize>(res: &T) -> Result<Response<Body>, Error> {
let resp_json = serde_json::to_string_pretty(res).map_err(garage_util::error::Error::from)?;
Ok(Response::builder()
.status(hyper::StatusCode::OK)
.header(http::header::CONTENT_TYPE, "application/json")
.body(Body::from(resp_json))?)
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn parse_bucket_containing_a_key() -> Result<(), Error> {
let (bucket, key) = parse_bucket_key("/my_bucket/a/super/file.jpg", None)?;
assert_eq!(bucket, "my_bucket");
assert_eq!(key.expect("key must be set"), "a/super/file.jpg");
Ok(())
}
#[test]
fn parse_bucket_containing_no_key() -> Result<(), Error> {
let (bucket, key) = parse_bucket_key("/my_bucket/", None)?;
assert_eq!(bucket, "my_bucket");
assert!(key.is_none());
let (bucket, key) = parse_bucket_key("/my_bucket", None)?;
assert_eq!(bucket, "my_bucket");
assert!(key.is_none());
Ok(())
}
#[test]
fn parse_bucket_containing_no_bucket() {
let parsed = parse_bucket_key("", None);
assert!(parsed.is_err());
let parsed = parse_bucket_key("/", None);
assert!(parsed.is_err());
let parsed = parse_bucket_key("////", None);
assert!(parsed.is_err());
}
#[test]
fn parse_bucket_with_vhost_and_key() -> Result<(), Error> {
let (bucket, key) = parse_bucket_key("/a/super/file.jpg", Some("my-bucket"))?;
assert_eq!(bucket, "my-bucket");
assert_eq!(key.expect("key must be set"), "a/super/file.jpg");
Ok(())
}
#[test]
fn parse_bucket_with_vhost_no_key() -> Result<(), Error> {
let (bucket, key) = parse_bucket_key("", Some("my-bucket"))?;
assert_eq!(bucket, "my-bucket");
assert!(key.is_none());
let (bucket, key) = parse_bucket_key("/", Some("my-bucket"))?;
assert_eq!(bucket, "my-bucket");
assert!(key.is_none());
Ok(())
}
#[test]
fn authority_to_host_with_port() -> Result<(), Error> {
let domain = authority_to_host("[::1]:3902")?;
@@ -111,4 +251,47 @@ mod tests {
assert_eq!(host_to_bucket("not-garage.tld", "garage.tld"), None);
assert_eq!(host_to_bucket("not-garage.tld", ".garage.tld"), None);
}
#[test]
fn test_key_after_prefix() {
use std::iter::FromIterator;
assert_eq!(UTF8_BEFORE_LAST_CHAR as u32, (char::MAX as u32) - 1);
assert_eq!(key_after_prefix("a/b/").unwrap().as_str(), "a/b0");
assert_eq!(key_after_prefix("").unwrap().as_str(), "");
assert_eq!(
key_after_prefix("􏿽").unwrap().as_str(),
String::from(char::from_u32(0x10FFFE).unwrap())
);
// When the last character is the biggest UTF8 char
let a = String::from_iter(['a', char::MAX].iter());
assert_eq!(key_after_prefix(a.as_str()).unwrap().as_str(), "b");
// When all characters are the biggest UTF8 char
let b = String::from_iter([char::MAX; 3].iter());
assert!(key_after_prefix(b.as_str()).is_none());
// Check utf8 surrogates
let c = String::from('\u{D7FF}');
assert_eq!(
key_after_prefix(c.as_str()).unwrap().as_str(),
String::from('\u{E000}')
);
// Check the character before the biggest one
let d = String::from('\u{10FFFE}');
assert_eq!(
key_after_prefix(d.as_str()).unwrap().as_str(),
String::from(char::MAX)
);
}
}
#[derive(Serialize)]
pub(crate) struct CustomApiErrorBody {
pub(crate) code: String,
pub(crate) message: String,
pub(crate) region: String,
pub(crate) path: String,
}
+190
View File
@@ -0,0 +1,190 @@
use std::net::SocketAddr;
use std::sync::Arc;
use async_trait::async_trait;
use futures::future::Future;
use hyper::{Body, Method, Request, Response};
use opentelemetry::{trace::SpanRef, KeyValue};
use garage_util::error::Error as GarageError;
use garage_model::garage::Garage;
use crate::generic_server::*;
use crate::k2v::error::*;
use crate::signature::payload::check_payload_signature;
use crate::signature::streaming::*;
use crate::helpers::*;
use crate::k2v::batch::*;
use crate::k2v::index::*;
use crate::k2v::item::*;
use crate::k2v::router::Endpoint;
use crate::s3::cors::*;
pub struct K2VApiServer {
garage: Arc<Garage>,
}
pub(crate) struct K2VApiEndpoint {
bucket_name: String,
endpoint: Endpoint,
}
impl K2VApiServer {
pub async fn run(
garage: Arc<Garage>,
bind_addr: SocketAddr,
s3_region: String,
shutdown_signal: impl Future<Output = ()>,
) -> Result<(), GarageError> {
ApiServer::new(s3_region, K2VApiServer { garage })
.run_server(bind_addr, shutdown_signal)
.await
}
}
#[async_trait]
impl ApiHandler for K2VApiServer {
const API_NAME: &'static str = "k2v";
const API_NAME_DISPLAY: &'static str = "K2V";
type Endpoint = K2VApiEndpoint;
type Error = Error;
fn parse_endpoint(&self, req: &Request<Body>) -> Result<K2VApiEndpoint, Error> {
let (endpoint, bucket_name) = Endpoint::from_request(req)?;
Ok(K2VApiEndpoint {
bucket_name,
endpoint,
})
}
async fn handle(
&self,
req: Request<Body>,
endpoint: K2VApiEndpoint,
) -> Result<Response<Body>, Error> {
let K2VApiEndpoint {
bucket_name,
endpoint,
} = endpoint;
let garage = self.garage.clone();
// The OPTIONS method is procesed early, before we even check for an API key
if let Endpoint::Options = endpoint {
return Ok(handle_options_s3api(garage, &req, Some(bucket_name))
.await
.ok_or_bad_request("Error handling OPTIONS")?);
}
let (api_key, mut content_sha256) = check_payload_signature(&garage, "k2v", &req).await?;
let api_key = api_key
.ok_or_else(|| Error::forbidden("Garage does not support anonymous access yet"))?;
let req = parse_streaming_body(
&api_key,
req,
&mut content_sha256,
&garage.config.s3_api.s3_region,
"k2v",
)?;
let bucket_id = garage
.bucket_helper()
.resolve_bucket(&bucket_name, &api_key)
.await?;
let bucket = garage
.bucket_helper()
.get_existing_bucket(bucket_id)
.await?;
let allowed = match endpoint.authorization_type() {
Authorization::Read => api_key.allow_read(&bucket_id),
Authorization::Write => api_key.allow_write(&bucket_id),
Authorization::Owner => api_key.allow_owner(&bucket_id),
_ => unreachable!(),
};
if !allowed {
return Err(Error::forbidden("Operation is not allowed for this key."));
}
// Look up what CORS rule might apply to response.
// Requests for methods different than GET, HEAD or POST
// are always preflighted, i.e. the browser should make
// an OPTIONS call before to check it is allowed
let matching_cors_rule = match *req.method() {
Method::GET | Method::HEAD | Method::POST => find_matching_cors_rule(&bucket, &req)
.ok_or_internal_error("Error looking up CORS rule")?,
_ => None,
};
let resp = match endpoint {
Endpoint::DeleteItem {
partition_key,
sort_key,
} => handle_delete_item(garage, req, bucket_id, &partition_key, &sort_key).await,
Endpoint::InsertItem {
partition_key,
sort_key,
} => handle_insert_item(garage, req, bucket_id, &partition_key, &sort_key).await,
Endpoint::ReadItem {
partition_key,
sort_key,
} => handle_read_item(garage, &req, bucket_id, &partition_key, &sort_key).await,
Endpoint::PollItem {
partition_key,
sort_key,
causality_token,
timeout,
} => {
handle_poll_item(
garage,
&req,
bucket_id,
partition_key,
sort_key,
causality_token,
timeout,
)
.await
}
Endpoint::ReadIndex {
prefix,
start,
end,
limit,
reverse,
} => handle_read_index(garage, bucket_id, prefix, start, end, limit, reverse).await,
Endpoint::InsertBatch {} => handle_insert_batch(garage, bucket_id, req).await,
Endpoint::ReadBatch {} => handle_read_batch(garage, bucket_id, req).await,
Endpoint::DeleteBatch {} => handle_delete_batch(garage, bucket_id, req).await,
Endpoint::Options => unreachable!(),
};
// If request was a success and we have a CORS rule that applies to it,
// add the corresponding CORS headers to the response
let mut resp_ok = resp?;
if let Some(rule) = matching_cors_rule {
add_cors_headers(&mut resp_ok, rule)
.ok_or_internal_error("Invalid bucket CORS configuration")?;
}
Ok(resp_ok)
}
}
impl ApiEndpoint for K2VApiEndpoint {
fn name(&self) -> &'static str {
self.endpoint.name()
}
fn add_span_attributes(&self, span: SpanRef<'_>) {
span.set_attribute(KeyValue::new("bucket", self.bucket_name.clone()));
}
}
+363
View File
@@ -0,0 +1,363 @@
use std::sync::Arc;
use hyper::{Body, Request, Response, StatusCode};
use serde::{Deserialize, Serialize};
use garage_util::data::*;
use garage_util::error::Error as GarageError;
use garage_table::{EnumerationOrder, TableSchema};
use garage_model::garage::Garage;
use garage_model::k2v::causality::*;
use garage_model::k2v::item_table::*;
use crate::helpers::*;
use crate::k2v::error::*;
use crate::k2v::range::read_range;
pub async fn handle_insert_batch(
garage: Arc<Garage>,
bucket_id: Uuid,
req: Request<Body>,
) -> Result<Response<Body>, Error> {
let items = parse_json_body::<Vec<InsertBatchItem>>(req).await?;
let mut items2 = vec![];
for it in items {
let ct = it
.ct
.map(|s| CausalContext::parse(&s))
.transpose()
.ok_or_bad_request("Invalid causality token")?;
let v = match it.v {
Some(vs) => {
DvvsValue::Value(base64::decode(vs).ok_or_bad_request("Invalid base64 value")?)
}
None => DvvsValue::Deleted,
};
items2.push((it.pk, it.sk, ct, v));
}
garage.k2v.rpc.insert_batch(bucket_id, items2).await?;
Ok(Response::builder()
.status(StatusCode::OK)
.body(Body::empty())?)
}
pub async fn handle_read_batch(
garage: Arc<Garage>,
bucket_id: Uuid,
req: Request<Body>,
) -> Result<Response<Body>, Error> {
let queries = parse_json_body::<Vec<ReadBatchQuery>>(req).await?;
let resp_results = futures::future::join_all(
queries
.into_iter()
.map(|q| handle_read_batch_query(&garage, bucket_id, q)),
)
.await;
let mut resps: Vec<ReadBatchResponse> = vec![];
for resp in resp_results {
resps.push(resp?);
}
let resp_json = serde_json::to_string_pretty(&resps).map_err(GarageError::from)?;
Ok(Response::builder()
.status(StatusCode::OK)
.body(Body::from(resp_json))?)
}
async fn handle_read_batch_query(
garage: &Arc<Garage>,
bucket_id: Uuid,
query: ReadBatchQuery,
) -> Result<ReadBatchResponse, Error> {
let partition = K2VItemPartition {
bucket_id,
partition_key: query.partition_key.clone(),
};
let filter = ItemFilter {
exclude_only_tombstones: !query.tombstones,
conflicts_only: query.conflicts_only,
};
let (items, more, next_start) = if query.single_item {
if query.prefix.is_some() || query.end.is_some() || query.limit.is_some() || query.reverse {
return Err(Error::bad_request("Batch query parameters 'prefix', 'end', 'limit' and 'reverse' must not be set when singleItem is true."));
}
let sk = query
.start
.as_ref()
.ok_or_bad_request("start should be specified if single_item is set")?;
let item = garage
.k2v
.item_table
.get(&partition, sk)
.await?
.filter(|e| K2VItemTable::matches_filter(e, &filter));
match item {
Some(i) => (vec![ReadBatchResponseItem::from(i)], false, None),
None => (vec![], false, None),
}
} else {
let (items, more, next_start) = read_range(
&garage.k2v.item_table,
&partition,
&query.prefix,
&query.start,
&query.end,
query.limit,
Some(filter),
EnumerationOrder::from_reverse(query.reverse),
)
.await?;
let items = items
.into_iter()
.map(ReadBatchResponseItem::from)
.collect::<Vec<_>>();
(items, more, next_start)
};
Ok(ReadBatchResponse {
partition_key: query.partition_key,
prefix: query.prefix,
start: query.start,
end: query.end,
limit: query.limit,
reverse: query.reverse,
single_item: query.single_item,
conflicts_only: query.conflicts_only,
tombstones: query.tombstones,
items,
more,
next_start,
})
}
pub async fn handle_delete_batch(
garage: Arc<Garage>,
bucket_id: Uuid,
req: Request<Body>,
) -> Result<Response<Body>, Error> {
let queries = parse_json_body::<Vec<DeleteBatchQuery>>(req).await?;
let resp_results = futures::future::join_all(
queries
.into_iter()
.map(|q| handle_delete_batch_query(&garage, bucket_id, q)),
)
.await;
let mut resps: Vec<DeleteBatchResponse> = vec![];
for resp in resp_results {
resps.push(resp?);
}
let resp_json = serde_json::to_string_pretty(&resps).map_err(GarageError::from)?;
Ok(Response::builder()
.status(StatusCode::OK)
.body(Body::from(resp_json))?)
}
async fn handle_delete_batch_query(
garage: &Arc<Garage>,
bucket_id: Uuid,
query: DeleteBatchQuery,
) -> Result<DeleteBatchResponse, Error> {
let partition = K2VItemPartition {
bucket_id,
partition_key: query.partition_key.clone(),
};
let filter = ItemFilter {
exclude_only_tombstones: true,
conflicts_only: false,
};
let deleted_items = if query.single_item {
if query.prefix.is_some() || query.end.is_some() {
return Err(Error::bad_request("Batch query parameters 'prefix' and 'end' must not be set when singleItem is true."));
}
let sk = query
.start
.as_ref()
.ok_or_bad_request("start should be specified if single_item is set")?;
let item = garage
.k2v
.item_table
.get(&partition, sk)
.await?
.filter(|e| K2VItemTable::matches_filter(e, &filter));
match item {
Some(i) => {
let cc = i.causal_context();
garage
.k2v
.rpc
.insert(
bucket_id,
i.partition.partition_key,
i.sort_key,
Some(cc),
DvvsValue::Deleted,
)
.await?;
1
}
None => 0,
}
} else {
let (items, more, _next_start) = read_range(
&garage.k2v.item_table,
&partition,
&query.prefix,
&query.start,
&query.end,
None,
Some(filter),
EnumerationOrder::Forward,
)
.await?;
assert!(!more);
// TODO delete items
let items = items
.into_iter()
.map(|i| {
let cc = i.causal_context();
(
i.partition.partition_key,
i.sort_key,
Some(cc),
DvvsValue::Deleted,
)
})
.collect::<Vec<_>>();
let n = items.len();
garage.k2v.rpc.insert_batch(bucket_id, items).await?;
n
};
Ok(DeleteBatchResponse {
partition_key: query.partition_key,
prefix: query.prefix,
start: query.start,
end: query.end,
single_item: query.single_item,
deleted_items,
})
}
#[derive(Deserialize)]
struct InsertBatchItem {
pk: String,
sk: String,
ct: Option<String>,
v: Option<String>,
}
#[derive(Deserialize)]
struct ReadBatchQuery {
#[serde(rename = "partitionKey")]
partition_key: String,
#[serde(default)]
prefix: Option<String>,
#[serde(default)]
start: Option<String>,
#[serde(default)]
end: Option<String>,
#[serde(default)]
limit: Option<u64>,
#[serde(default)]
reverse: bool,
#[serde(default, rename = "singleItem")]
single_item: bool,
#[serde(default, rename = "conflictsOnly")]
conflicts_only: bool,
#[serde(default)]
tombstones: bool,
}
#[derive(Serialize)]
struct ReadBatchResponse {
#[serde(rename = "partitionKey")]
partition_key: String,
prefix: Option<String>,
start: Option<String>,
end: Option<String>,
limit: Option<u64>,
reverse: bool,
#[serde(rename = "singleItem")]
single_item: bool,
#[serde(rename = "conflictsOnly")]
conflicts_only: bool,
tombstones: bool,
items: Vec<ReadBatchResponseItem>,
more: bool,
#[serde(rename = "nextStart")]
next_start: Option<String>,
}
#[derive(Serialize)]
struct ReadBatchResponseItem {
sk: String,
ct: String,
v: Vec<Option<String>>,
}
impl ReadBatchResponseItem {
fn from(i: K2VItem) -> Self {
let ct = i.causal_context().serialize();
let v = i
.values()
.iter()
.map(|v| match v {
DvvsValue::Value(x) => Some(base64::encode(x)),
DvvsValue::Deleted => None,
})
.collect::<Vec<_>>();
Self {
sk: i.sort_key,
ct,
v,
}
}
}
#[derive(Deserialize)]
struct DeleteBatchQuery {
#[serde(rename = "partitionKey")]
partition_key: String,
#[serde(default)]
prefix: Option<String>,
#[serde(default)]
start: Option<String>,
#[serde(default)]
end: Option<String>,
#[serde(default, rename = "singleItem")]
single_item: bool,
}
#[derive(Serialize)]
struct DeleteBatchResponse {
#[serde(rename = "partitionKey")]
partition_key: String,
prefix: Option<String>,
start: Option<String>,
end: Option<String>,
#[serde(rename = "singleItem")]
single_item: bool,
#[serde(rename = "deletedItems")]
deleted_items: usize,
}
+135
View File
@@ -0,0 +1,135 @@
use err_derive::Error;
use hyper::header::HeaderValue;
use hyper::{Body, HeaderMap, StatusCode};
use garage_model::helper::error::Error as HelperError;
use crate::common_error::CommonError;
pub use crate::common_error::{CommonErrorDerivative, OkOrBadRequest, OkOrInternalError};
use crate::generic_server::ApiError;
use crate::helpers::CustomApiErrorBody;
use crate::signature::error::Error as SignatureError;
/// Errors of this crate
#[derive(Debug, Error)]
pub enum Error {
#[error(display = "{}", _0)]
/// Error from common error
Common(CommonError),
// Category: cannot process
/// Authorization Header Malformed
#[error(display = "Authorization header malformed, expected scope: {}", _0)]
AuthorizationHeaderMalformed(String),
/// The object requested don't exists
#[error(display = "Key not found")]
NoSuchKey,
/// Some base64 encoded data was badly encoded
#[error(display = "Invalid base64: {}", _0)]
InvalidBase64(#[error(source)] base64::DecodeError),
/// The client sent a header with invalid value
#[error(display = "Invalid header value: {}", _0)]
InvalidHeader(#[error(source)] hyper::header::ToStrError),
/// The client asked for an invalid return format (invalid Accept header)
#[error(display = "Not acceptable: {}", _0)]
NotAcceptable(String),
/// The request contained an invalid UTF-8 sequence in its path or in other parameters
#[error(display = "Invalid UTF-8: {}", _0)]
InvalidUtf8Str(#[error(source)] std::str::Utf8Error),
}
impl<T> From<T> for Error
where
CommonError: From<T>,
{
fn from(err: T) -> Self {
Error::Common(CommonError::from(err))
}
}
impl CommonErrorDerivative for Error {}
impl From<HelperError> for Error {
fn from(err: HelperError) -> Self {
match err {
HelperError::Internal(i) => Self::Common(CommonError::InternalError(i)),
HelperError::BadRequest(b) => Self::Common(CommonError::BadRequest(b)),
HelperError::InvalidBucketName(n) => Self::Common(CommonError::InvalidBucketName(n)),
HelperError::NoSuchBucket(n) => Self::Common(CommonError::NoSuchBucket(n)),
e => Self::Common(CommonError::BadRequest(format!("{}", e))),
}
}
}
impl From<SignatureError> for Error {
fn from(err: SignatureError) -> Self {
match err {
SignatureError::Common(c) => Self::Common(c),
SignatureError::AuthorizationHeaderMalformed(c) => {
Self::AuthorizationHeaderMalformed(c)
}
SignatureError::InvalidUtf8Str(i) => Self::InvalidUtf8Str(i),
SignatureError::InvalidHeader(h) => Self::InvalidHeader(h),
}
}
}
impl Error {
/// This returns a keyword for the corresponding error.
/// Here, these keywords are not necessarily those from AWS S3,
/// as we are building a custom API
fn code(&self) -> &'static str {
match self {
Error::Common(c) => c.aws_code(),
Error::NoSuchKey => "NoSuchKey",
Error::NotAcceptable(_) => "NotAcceptable",
Error::AuthorizationHeaderMalformed(_) => "AuthorizationHeaderMalformed",
Error::InvalidBase64(_) => "InvalidBase64",
Error::InvalidHeader(_) => "InvalidHeaderValue",
Error::InvalidUtf8Str(_) => "InvalidUtf8String",
}
}
}
impl ApiError for Error {
/// Get the HTTP status code that best represents the meaning of the error for the client
fn http_status_code(&self) -> StatusCode {
match self {
Error::Common(c) => c.http_status_code(),
Error::NoSuchKey => StatusCode::NOT_FOUND,
Error::NotAcceptable(_) => StatusCode::NOT_ACCEPTABLE,
Error::AuthorizationHeaderMalformed(_)
| Error::InvalidBase64(_)
| Error::InvalidHeader(_)
| Error::InvalidUtf8Str(_) => StatusCode::BAD_REQUEST,
}
}
fn add_http_headers(&self, header_map: &mut HeaderMap<HeaderValue>) {
use hyper::header;
header_map.append(header::CONTENT_TYPE, "application/json".parse().unwrap());
}
fn http_body(&self, garage_region: &str, path: &str) -> Body {
let error = CustomApiErrorBody {
code: self.code().to_string(),
message: format!("{}", self),
path: path.to_string(),
region: garage_region.to_string(),
};
Body::from(serde_json::to_string_pretty(&error).unwrap_or_else(|_| {
r#"
{
"code": "InternalError",
"message": "JSON encoding of error failed"
}
"#
.into()
}))
}
}
+100
View File
@@ -0,0 +1,100 @@
use std::sync::Arc;
use hyper::{Body, Response, StatusCode};
use serde::Serialize;
use garage_util::data::*;
use garage_util::error::Error as GarageError;
use garage_rpc::ring::Ring;
use garage_table::util::*;
use garage_model::garage::Garage;
use garage_model::k2v::item_table::{BYTES, CONFLICTS, ENTRIES, VALUES};
use crate::k2v::error::*;
use crate::k2v::range::read_range;
pub async fn handle_read_index(
garage: Arc<Garage>,
bucket_id: Uuid,
prefix: Option<String>,
start: Option<String>,
end: Option<String>,
limit: Option<u64>,
reverse: Option<bool>,
) -> Result<Response<Body>, Error> {
let reverse = reverse.unwrap_or(false);
let ring: Arc<Ring> = garage.system.ring.borrow().clone();
let (partition_keys, more, next_start) = read_range(
&garage.k2v.counter_table.table,
&bucket_id,
&prefix,
&start,
&end,
limit,
Some((DeletedFilter::NotDeleted, ring.layout.node_id_vec.clone())),
EnumerationOrder::from_reverse(reverse),
)
.await?;
let s_entries = ENTRIES.to_string();
let s_conflicts = CONFLICTS.to_string();
let s_values = VALUES.to_string();
let s_bytes = BYTES.to_string();
let resp = ReadIndexResponse {
prefix,
start,
end,
limit,
reverse,
partition_keys: partition_keys
.into_iter()
.map(|part| {
let vals = part.filtered_values(&ring);
ReadIndexResponseEntry {
pk: part.sk,
entries: *vals.get(&s_entries).unwrap_or(&0),
conflicts: *vals.get(&s_conflicts).unwrap_or(&0),
values: *vals.get(&s_values).unwrap_or(&0),
bytes: *vals.get(&s_bytes).unwrap_or(&0),
}
})
.collect::<Vec<_>>(),
more,
next_start,
};
let resp_json = serde_json::to_string_pretty(&resp).map_err(GarageError::from)?;
Ok(Response::builder()
.status(StatusCode::OK)
.body(Body::from(resp_json))?)
}
#[derive(Serialize)]
struct ReadIndexResponse {
prefix: Option<String>,
start: Option<String>,
end: Option<String>,
limit: Option<u64>,
reverse: bool,
#[serde(rename = "partitionKeys")]
partition_keys: Vec<ReadIndexResponseEntry>,
more: bool,
#[serde(rename = "nextStart")]
next_start: Option<String>,
}
#[derive(Serialize)]
struct ReadIndexResponseEntry {
pk: String,
entries: i64,
conflicts: i64,
values: i64,
bytes: i64,
}
+230
View File
@@ -0,0 +1,230 @@
use std::sync::Arc;
use http::header;
use hyper::{Body, Request, Response, StatusCode};
use garage_util::data::*;
use garage_model::garage::Garage;
use garage_model::k2v::causality::*;
use garage_model::k2v::item_table::*;
use crate::k2v::error::*;
pub const X_GARAGE_CAUSALITY_TOKEN: &str = "X-Garage-Causality-Token";
pub enum ReturnFormat {
Json,
Binary,
Either,
}
impl ReturnFormat {
pub fn from(req: &Request<Body>) -> Result<Self, Error> {
let accept = match req.headers().get(header::ACCEPT) {
Some(a) => a.to_str()?,
None => return Ok(Self::Json),
};
let accept = accept.split(',').map(|s| s.trim()).collect::<Vec<_>>();
let accept_json = accept.contains(&"application/json") || accept.contains(&"*/*");
let accept_binary = accept.contains(&"application/octet-stream") || accept.contains(&"*/*");
match (accept_json, accept_binary) {
(true, true) => Ok(Self::Either),
(true, false) => Ok(Self::Json),
(false, true) => Ok(Self::Binary),
(false, false) => Err(Error::NotAcceptable("Invalid Accept: header value, must contain either application/json or application/octet-stream (or both)".into())),
}
}
pub fn make_response(&self, item: &K2VItem) -> Result<Response<Body>, Error> {
let vals = item.values();
if vals.is_empty() {
return Err(Error::NoSuchKey);
}
let ct = item.causal_context().serialize();
match self {
Self::Binary if vals.len() > 1 => Ok(Response::builder()
.header(X_GARAGE_CAUSALITY_TOKEN, ct)
.status(StatusCode::CONFLICT)
.body(Body::empty())?),
Self::Binary => {
assert!(vals.len() == 1);
Self::make_binary_response(ct, vals[0])
}
Self::Either if vals.len() == 1 => Self::make_binary_response(ct, vals[0]),
_ => Self::make_json_response(ct, &vals[..]),
}
}
fn make_binary_response(ct: String, v: &DvvsValue) -> Result<Response<Body>, Error> {
match v {
DvvsValue::Deleted => Ok(Response::builder()
.header(X_GARAGE_CAUSALITY_TOKEN, ct)
.header(header::CONTENT_TYPE, "application/octet-stream")
.status(StatusCode::NO_CONTENT)
.body(Body::empty())?),
DvvsValue::Value(v) => Ok(Response::builder()
.header(X_GARAGE_CAUSALITY_TOKEN, ct)
.header(header::CONTENT_TYPE, "application/octet-stream")
.status(StatusCode::OK)
.body(Body::from(v.to_vec()))?),
}
}
fn make_json_response(ct: String, v: &[&DvvsValue]) -> Result<Response<Body>, Error> {
let items = v
.iter()
.map(|v| match v {
DvvsValue::Deleted => serde_json::Value::Null,
DvvsValue::Value(v) => serde_json::Value::String(base64::encode(v)),
})
.collect::<Vec<_>>();
let json_body =
serde_json::to_string_pretty(&items).ok_or_internal_error("JSON encoding error")?;
Ok(Response::builder()
.header(X_GARAGE_CAUSALITY_TOKEN, ct)
.header(header::CONTENT_TYPE, "application/json")
.status(StatusCode::OK)
.body(Body::from(json_body))?)
}
}
/// Handle ReadItem request
#[allow(clippy::ptr_arg)]
pub async fn handle_read_item(
garage: Arc<Garage>,
req: &Request<Body>,
bucket_id: Uuid,
partition_key: &str,
sort_key: &String,
) -> Result<Response<Body>, Error> {
let format = ReturnFormat::from(req)?;
let item = garage
.k2v
.item_table
.get(
&K2VItemPartition {
bucket_id,
partition_key: partition_key.to_string(),
},
sort_key,
)
.await?
.ok_or(Error::NoSuchKey)?;
format.make_response(&item)
}
pub async fn handle_insert_item(
garage: Arc<Garage>,
req: Request<Body>,
bucket_id: Uuid,
partition_key: &str,
sort_key: &str,
) -> Result<Response<Body>, Error> {
let causal_context = req
.headers()
.get(X_GARAGE_CAUSALITY_TOKEN)
.map(|s| s.to_str())
.transpose()?
.map(CausalContext::parse)
.transpose()
.ok_or_bad_request("Invalid causality token")?;
let body = hyper::body::to_bytes(req.into_body()).await?;
let value = DvvsValue::Value(body.to_vec());
garage
.k2v
.rpc
.insert(
bucket_id,
partition_key.to_string(),
sort_key.to_string(),
causal_context,
value,
)
.await?;
Ok(Response::builder()
.status(StatusCode::OK)
.body(Body::empty())?)
}
pub async fn handle_delete_item(
garage: Arc<Garage>,
req: Request<Body>,
bucket_id: Uuid,
partition_key: &str,
sort_key: &str,
) -> Result<Response<Body>, Error> {
let causal_context = req
.headers()
.get(X_GARAGE_CAUSALITY_TOKEN)
.map(|s| s.to_str())
.transpose()?
.map(CausalContext::parse)
.transpose()
.ok_or_bad_request("Invalid causality token")?;
let value = DvvsValue::Deleted;
garage
.k2v
.rpc
.insert(
bucket_id,
partition_key.to_string(),
sort_key.to_string(),
causal_context,
value,
)
.await?;
Ok(Response::builder()
.status(StatusCode::NO_CONTENT)
.body(Body::empty())?)
}
/// Handle ReadItem request
#[allow(clippy::ptr_arg)]
pub async fn handle_poll_item(
garage: Arc<Garage>,
req: &Request<Body>,
bucket_id: Uuid,
partition_key: String,
sort_key: String,
causality_token: String,
timeout_secs: Option<u64>,
) -> Result<Response<Body>, Error> {
let format = ReturnFormat::from(req)?;
let causal_context =
CausalContext::parse(&causality_token).ok_or_bad_request("Invalid causality token")?;
let item = garage
.k2v
.rpc
.poll(
bucket_id,
partition_key,
sort_key,
causal_context,
timeout_secs.unwrap_or(300) * 1000,
)
.await?;
if let Some(item) = item {
format.make_response(&item)
} else {
Ok(Response::builder()
.status(StatusCode::NOT_MODIFIED)
.body(Body::empty())?)
}
}
+9
View File
@@ -0,0 +1,9 @@
pub mod api_server;
mod error;
mod router;
mod batch;
mod index;
mod item;
mod range;
+100
View File
@@ -0,0 +1,100 @@
//! Utility module for retrieving ranges of items in Garage tables
//! Implements parameters (prefix, start, end, limit) as specified
//! for endpoints ReadIndex, ReadBatch and DeleteBatch
use std::sync::Arc;
use garage_table::replication::TableShardedReplication;
use garage_table::*;
use crate::helpers::key_after_prefix;
use crate::k2v::error::*;
/// Read range in a Garage table.
/// Returns (entries, more?, nextStart)
#[allow(clippy::too_many_arguments)]
pub(crate) async fn read_range<F>(
table: &Arc<Table<F, TableShardedReplication>>,
partition_key: &F::P,
prefix: &Option<String>,
start: &Option<String>,
end: &Option<String>,
limit: Option<u64>,
filter: Option<F::Filter>,
enumeration_order: EnumerationOrder,
) -> Result<(Vec<F::E>, bool, Option<String>), Error>
where
F: TableSchema<S = String> + 'static,
{
let (mut start, mut start_ignore) = match (prefix, start) {
(None, None) => (None, false),
(None, Some(s)) => (Some(s.clone()), false),
(Some(p), Some(s)) => {
if !s.starts_with(p) {
return Err(Error::bad_request(format!(
"Start key '{}' does not start with prefix '{}'",
s, p
)));
}
(Some(s.clone()), false)
}
(Some(p), None) if enumeration_order == EnumerationOrder::Reverse => {
let start = key_after_prefix(p)
.ok_or_internal_error("Sorry, can't list this prefix in reverse order")?;
(Some(start), true)
}
(Some(p), None) => (Some(p.clone()), false),
};
let mut entries = vec![];
loop {
let n_get = std::cmp::min(
1000,
limit.map(|x| x as usize).unwrap_or(usize::MAX - 10) - entries.len() + 2,
);
let get_ret = table
.get_range(
partition_key,
start.clone(),
filter.clone(),
n_get,
enumeration_order,
)
.await?;
let get_ret_len = get_ret.len();
for entry in get_ret {
if start_ignore && Some(entry.sort_key()) == start.as_ref() {
continue;
}
if let Some(p) = prefix {
if !entry.sort_key().starts_with(p) {
return Ok((entries, false, None));
}
}
if let Some(e) = end {
let is_finished = match enumeration_order {
EnumerationOrder::Forward => entry.sort_key() >= e,
EnumerationOrder::Reverse => entry.sort_key() <= e,
};
if is_finished {
return Ok((entries, false, None));
}
}
if let Some(l) = limit {
if entries.len() >= l as usize {
return Ok((entries, true, Some(entry.sort_key().clone())));
}
}
entries.push(entry);
}
if get_ret_len < n_get {
return Ok((entries, false, None));
}
start = Some(entries.last().unwrap().sort_key().clone());
start_ignore = true;
}
}
+252
View File
@@ -0,0 +1,252 @@
use crate::k2v::error::*;
use std::borrow::Cow;
use hyper::{Method, Request};
use crate::helpers::Authorization;
use crate::router_macros::{generateQueryParameters, router_match};
router_match! {@func
/// List of all K2V API endpoints.
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Endpoint {
DeleteBatch {
},
DeleteItem {
partition_key: String,
sort_key: String,
},
InsertBatch {
},
InsertItem {
partition_key: String,
sort_key: String,
},
Options,
PollItem {
partition_key: String,
sort_key: String,
causality_token: String,
timeout: Option<u64>,
},
ReadBatch {
},
ReadIndex {
prefix: Option<String>,
start: Option<String>,
end: Option<String>,
limit: Option<u64>,
reverse: Option<bool>,
},
ReadItem {
partition_key: String,
sort_key: String,
},
}}
impl Endpoint {
/// Determine which S3 endpoint a request is for using the request, and a bucket which was
/// possibly extracted from the Host header.
/// Returns Self plus bucket name, if endpoint is not Endpoint::ListBuckets
pub fn from_request<T>(req: &Request<T>) -> Result<(Self, String), Error> {
let uri = req.uri();
let path = uri.path().trim_start_matches('/');
let query = uri.query();
let (bucket, partition_key) = path
.split_once('/')
.map(|(b, p)| (b.to_owned(), p.trim_start_matches('/')))
.unwrap_or((path.to_owned(), ""));
if bucket.is_empty() {
return Err(Error::bad_request("Missing bucket name"));
}
if *req.method() == Method::OPTIONS {
return Ok((Self::Options, bucket));
}
let partition_key = percent_encoding::percent_decode_str(partition_key)
.decode_utf8()?
.into_owned();
let mut query = QueryParameters::from_query(query.unwrap_or_default())?;
let method_search = Method::from_bytes(b"SEARCH").unwrap();
let res = match *req.method() {
Method::GET => Self::from_get(partition_key, &mut query)?,
//&Method::HEAD => Self::from_head(partition_key, &mut query)?,
Method::POST => Self::from_post(partition_key, &mut query)?,
Method::PUT => Self::from_put(partition_key, &mut query)?,
Method::DELETE => Self::from_delete(partition_key, &mut query)?,
_ if req.method() == method_search => Self::from_search(partition_key, &mut query)?,
_ => return Err(Error::bad_request("Unknown method")),
};
if let Some(message) = query.nonempty_message() {
debug!("Unused query parameter: {}", message)
}
Ok((res, bucket))
}
/// Determine which endpoint a request is for, knowing it is a GET.
fn from_get(partition_key: String, query: &mut QueryParameters<'_>) -> Result<Self, Error> {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), partition_key, query, None),
key: [
EMPTY if causality_token => PollItem (query::sort_key, query::causality_token, opt_parse::timeout),
EMPTY => ReadItem (query::sort_key),
],
no_key: [
EMPTY => ReadIndex (query_opt::prefix, query_opt::start, query_opt::end, opt_parse::limit, opt_parse::reverse),
]
}
}
/// Determine which endpoint a request is for, knowing it is a SEARCH.
fn from_search(partition_key: String, query: &mut QueryParameters<'_>) -> Result<Self, Error> {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), partition_key, query, None),
key: [
],
no_key: [
EMPTY => ReadBatch,
]
}
}
/*
/// Determine which endpoint a request is for, knowing it is a HEAD.
fn from_head(partition_key: String, query: &mut QueryParameters<'_>) -> Result<Self, Error> {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), partition_key, query, None),
key: [
EMPTY => HeadObject(opt_parse::part_number, query_opt::version_id),
],
no_key: [
EMPTY => HeadBucket,
]
}
}
*/
/// Determine which endpoint a request is for, knowing it is a POST.
fn from_post(partition_key: String, query: &mut QueryParameters<'_>) -> Result<Self, Error> {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), partition_key, query, None),
key: [
],
no_key: [
EMPTY => InsertBatch,
DELETE => DeleteBatch,
SEARCH => ReadBatch,
]
}
}
/// Determine which endpoint a request is for, knowing it is a PUT.
fn from_put(partition_key: String, query: &mut QueryParameters<'_>) -> Result<Self, Error> {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), partition_key, query, None),
key: [
EMPTY => InsertItem (query::sort_key),
],
no_key: [
]
}
}
/// Determine which endpoint a request is for, knowing it is a DELETE.
fn from_delete(partition_key: String, query: &mut QueryParameters<'_>) -> Result<Self, Error> {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), partition_key, query, None),
key: [
EMPTY => DeleteItem (query::sort_key),
],
no_key: [
]
}
}
/// Get the partition key the request target. Returns None for requests which don't use a partition key.
#[allow(dead_code)]
pub fn get_partition_key(&self) -> Option<&str> {
router_match! {
@extract
self,
partition_key,
[
DeleteItem,
InsertItem,
PollItem,
ReadItem,
]
}
}
/// Get the sort key the request target. Returns None for requests which don't use a sort key.
#[allow(dead_code)]
pub fn get_sort_key(&self) -> Option<&str> {
router_match! {
@extract
self,
sort_key,
[
DeleteItem,
InsertItem,
PollItem,
ReadItem,
]
}
}
/// Get the kind of authorization which is required to perform the operation.
pub fn authorization_type(&self) -> Authorization {
let readonly = router_match! {
@match
self,
[
PollItem,
ReadBatch,
ReadIndex,
ReadItem,
]
};
if readonly {
Authorization::Read
} else {
Authorization::Write
}
}
}
// parameter name => struct field
generateQueryParameters! {
"prefix" => prefix,
"start" => start,
"causality_token" => causality_token,
"end" => end,
"limit" => limit,
"reverse" => reverse,
"sort_key" => sort_key,
"timeout" => timeout
}
mod keywords {
//! This module contain all query parameters with no associated value
//! used to differentiate endpoints.
pub const EMPTY: &str = "";
pub const DELETE: &str = "delete";
pub const SEARCH: &str = "search";
}
+8 -18
View File
@@ -2,26 +2,16 @@
#[macro_use]
extern crate tracing;
pub mod error;
pub use error::Error;
pub mod common_error;
mod encoding;
mod api_server;
pub use api_server::run_api_server;
pub mod generic_server;
pub mod helpers;
mod router_macros;
/// This mode is public only to help testing. Don't expect stability here
pub mod signature;
pub mod helpers;
mod s3_bucket;
mod s3_copy;
pub mod s3_cors;
mod s3_delete;
pub mod s3_get;
mod s3_list;
mod s3_post_object;
mod s3_put;
mod s3_router;
mod s3_website;
mod s3_xml;
pub mod admin;
#[cfg(feature = "k2v")]
pub mod k2v;
pub mod s3;
+213
View File
@@ -0,0 +1,213 @@
/// This macro is used to generate very repetitive match {} blocks in this module
/// It is _not_ made to be used anywhere else
macro_rules! router_match {
(@match $enum:expr , [ $($endpoint:ident,)* ]) => {{
// usage: router_match {@match my_enum, [ VariantWithField1, VariantWithField2 ..] }
// returns true if the variant was one of the listed variants, false otherwise.
use Endpoint::*;
match $enum {
$(
$endpoint { .. } => true,
)*
_ => false
}
}};
(@extract $enum:expr , $param:ident, [ $($endpoint:ident,)* ]) => {{
// usage: router_match {@extract my_enum, field_name, [ VariantWithField1, VariantWithField2 ..] }
// returns Some(field_value), or None if the variant was not one of the listed variants.
use Endpoint::*;
match $enum {
$(
$endpoint {$param, ..} => Some($param),
)*
_ => None
}
}};
(@gen_path_parser ($method:expr, $reqpath:expr, $query:expr)
[
$($meth:ident $path:pat $(if $required:ident)? => $api:ident $(($($conv:ident :: $param:ident),*))?,)*
]) => {{
{
use Endpoint::*;
match ($method, $reqpath) {
$(
(&Method::$meth, $path) if true $(&& $query.$required.is_some())? => $api {
$($(
$param: router_match!(@@parse_param $query, $conv, $param),
)*)?
},
)*
(m, p) => {
return Err(Error::bad_request(format!(
"Unknown API endpoint: {} {}",
m, p
)))
}
}
}
}};
(@gen_parser ($keyword:expr, $key:ident, $query:expr, $header:expr),
key: [$($kw_k:ident $(if $required_k:ident)? $(header $header_k:expr)? => $api_k:ident $(($($conv_k:ident :: $param_k:ident),*))?,)*],
no_key: [$($kw_nk:ident $(if $required_nk:ident)? $(if_header $header_nk:expr)? => $api_nk:ident $(($($conv_nk:ident :: $param_nk:ident),*))?,)*]) => {{
// usage: router_match {@gen_parser (keyword, key, query, header),
// key: [
// SOME_KEYWORD => VariantWithKey,
// ...
// ],
// no_key: [
// SOME_KEYWORD => VariantWithoutKey,
// ...
// ]
// }
// See in from_{method} for more detailed usage.
use Endpoint::*;
use keywords::*;
match ($keyword, !$key.is_empty()){
$(
($kw_k, true) if true $(&& $query.$required_k.is_some())? $(&& $header.contains_key($header_k))? => Ok($api_k {
$key,
$($(
$param_k: router_match!(@@parse_param $query, $conv_k, $param_k),
)*)?
}),
)*
$(
($kw_nk, false) $(if $query.$required_nk.is_some())? $(if $header.contains($header_nk))? => Ok($api_nk {
$($(
$param_nk: router_match!(@@parse_param $query, $conv_nk, $param_nk),
)*)?
}),
)*
(kw, _) => Err(Error::bad_request(format!("Invalid endpoint: {}", kw)))
}
}};
(@@parse_param $query:expr, query_opt, $param:ident) => {{
// extract optional query parameter
$query.$param.take().map(|param| param.into_owned())
}};
(@@parse_param $query:expr, query, $param:ident) => {{
// extract mendatory query parameter
$query.$param.take().ok_or_bad_request("Missing argument for endpoint")?.into_owned()
}};
(@@parse_param $query:expr, opt_parse, $param:ident) => {{
// extract and parse optional query parameter
// missing parameter is file, however parse error is reported as an error
$query.$param
.take()
.map(|param| param.parse())
.transpose()
.map_err(|_| Error::bad_request("Failed to parse query parameter"))?
}};
(@@parse_param $query:expr, parse, $param:ident) => {{
// extract and parse mandatory query parameter
// both missing and un-parseable parameters are reported as errors
$query.$param.take().ok_or_bad_request("Missing argument for endpoint")?
.parse()
.map_err(|_| Error::bad_request("Failed to parse query parameter"))?
}};
(@func
$(#[$doc:meta])*
pub enum Endpoint {
$(
$(#[$outer:meta])*
$variant:ident $({
$($name:ident: $ty:ty,)*
})?,
)*
}) => {
$(#[$doc])*
pub enum Endpoint {
$(
$(#[$outer])*
$variant $({
$($name: $ty, )*
})?,
)*
}
impl Endpoint {
pub fn name(&self) -> &'static str {
match self {
$(Endpoint::$variant $({ $($name: _,)* .. })? => stringify!($variant),)*
}
}
}
};
(@if ($($cond:tt)+) then ($($then:tt)*) else ($($else:tt)*)) => {
$($then)*
};
(@if () then ($($then:tt)*) else ($($else:tt)*)) => {
$($else)*
};
}
/// This macro is used to generate part of the code in this module. It must be called only one, and
/// is useless outside of this module.
macro_rules! generateQueryParameters {
( $($rest:expr => $name:ident),* ) => {
/// Struct containing all query parameters used in endpoints. Think of it as an HashMap,
/// but with keys statically known.
#[derive(Debug, Default)]
struct QueryParameters<'a> {
keyword: Option<Cow<'a, str>>,
$(
$name: Option<Cow<'a, str>>,
)*
}
impl<'a> QueryParameters<'a> {
/// Build this struct from the query part of an URI.
fn from_query(query: &'a str) -> Result<Self, Error> {
let mut res: Self = Default::default();
for (k, v) in url::form_urlencoded::parse(query.as_bytes()) {
let repeated = match k.as_ref() {
$(
$rest => if !v.is_empty() {
res.$name.replace(v).is_some()
} else {
false
},
)*
_ => {
if k.starts_with("response-") || k.starts_with("X-Amz-") {
false
} else if v.as_ref().is_empty() {
if res.keyword.replace(k).is_some() {
return Err(Error::bad_request("Multiple keywords"));
}
continue;
} else {
debug!("Received an unknown query parameter: '{}'", k);
false
}
}
};
if repeated {
return Err(Error::bad_request(format!(
"Query parameter repeated: '{}'",
k
)));
}
}
Ok(res)
}
/// Get an error message in case not all parameters where used when extracting them to
/// build an Enpoint variant
fn nonempty_message(&self) -> Option<&str> {
if self.keyword.is_some() {
Some("Keyword not used")
} $(
else if self.$name.is_some() {
Some(concat!("'", $rest, "'"))
}
)* else {
None
}
}
}
}
}
pub(crate) use generateQueryParameters;
pub(crate) use router_match;
+390
View File
@@ -0,0 +1,390 @@
use std::net::SocketAddr;
use std::sync::Arc;
use async_trait::async_trait;
use futures::future::Future;
use hyper::header;
use hyper::{Body, Request, Response};
use opentelemetry::{trace::SpanRef, KeyValue};
use garage_util::error::Error as GarageError;
use garage_model::garage::Garage;
use garage_model::key_table::Key;
use crate::generic_server::*;
use crate::s3::error::*;
use crate::signature::payload::check_payload_signature;
use crate::signature::streaming::*;
use crate::helpers::*;
use crate::s3::bucket::*;
use crate::s3::copy::*;
use crate::s3::cors::*;
use crate::s3::delete::*;
use crate::s3::get::*;
use crate::s3::list::*;
use crate::s3::post_object::handle_post_object;
use crate::s3::put::*;
use crate::s3::router::Endpoint;
use crate::s3::website::*;
pub struct S3ApiServer {
garage: Arc<Garage>,
}
pub(crate) struct S3ApiEndpoint {
bucket_name: Option<String>,
endpoint: Endpoint,
}
impl S3ApiServer {
pub async fn run(
garage: Arc<Garage>,
addr: SocketAddr,
s3_region: String,
shutdown_signal: impl Future<Output = ()>,
) -> Result<(), GarageError> {
ApiServer::new(s3_region, S3ApiServer { garage })
.run_server(addr, shutdown_signal)
.await
}
async fn handle_request_without_bucket(
&self,
_req: Request<Body>,
api_key: Key,
endpoint: Endpoint,
) -> Result<Response<Body>, Error> {
match endpoint {
Endpoint::ListBuckets => handle_list_buckets(&self.garage, &api_key).await,
endpoint => Err(Error::NotImplemented(endpoint.name().to_owned())),
}
}
}
#[async_trait]
impl ApiHandler for S3ApiServer {
const API_NAME: &'static str = "s3";
const API_NAME_DISPLAY: &'static str = "S3";
type Endpoint = S3ApiEndpoint;
type Error = Error;
fn parse_endpoint(&self, req: &Request<Body>) -> Result<S3ApiEndpoint, Error> {
let authority = req
.headers()
.get(header::HOST)
.ok_or_bad_request("Host header required")?
.to_str()?;
let host = authority_to_host(authority)?;
let bucket_name = self
.garage
.config
.s3_api
.root_domain
.as_ref()
.and_then(|root_domain| host_to_bucket(&host, root_domain));
let (endpoint, bucket_name) =
Endpoint::from_request(req, bucket_name.map(ToOwned::to_owned))?;
Ok(S3ApiEndpoint {
bucket_name,
endpoint,
})
}
async fn handle(
&self,
req: Request<Body>,
endpoint: S3ApiEndpoint,
) -> Result<Response<Body>, Error> {
let S3ApiEndpoint {
bucket_name,
endpoint,
} = endpoint;
let garage = self.garage.clone();
// Some endpoints are processed early, before we even check for an API key
if let Endpoint::PostObject = endpoint {
return handle_post_object(garage, req, bucket_name.unwrap()).await;
}
if let Endpoint::Options = endpoint {
return handle_options_s3api(garage, &req, bucket_name).await;
}
let (api_key, mut content_sha256) = check_payload_signature(&garage, "s3", &req).await?;
let api_key = api_key
.ok_or_else(|| Error::forbidden("Garage does not support anonymous access yet"))?;
let req = parse_streaming_body(
&api_key,
req,
&mut content_sha256,
&garage.config.s3_api.s3_region,
"s3",
)?;
let bucket_name = match bucket_name {
None => {
return self
.handle_request_without_bucket(req, api_key, endpoint)
.await
}
Some(bucket) => bucket.to_string(),
};
// Special code path for CreateBucket API endpoint
if let Endpoint::CreateBucket {} = endpoint {
return handle_create_bucket(&garage, req, content_sha256, api_key, bucket_name).await;
}
let bucket_id = garage
.bucket_helper()
.resolve_bucket(&bucket_name, &api_key)
.await?;
let bucket = garage
.bucket_helper()
.get_existing_bucket(bucket_id)
.await?;
let allowed = match endpoint.authorization_type() {
Authorization::Read => api_key.allow_read(&bucket_id),
Authorization::Write => api_key.allow_write(&bucket_id),
Authorization::Owner => api_key.allow_owner(&bucket_id),
_ => unreachable!(),
};
if !allowed {
return Err(Error::forbidden("Operation is not allowed for this key."));
}
let matching_cors_rule = find_matching_cors_rule(&bucket, &req)?;
let resp = match endpoint {
Endpoint::HeadObject {
key, part_number, ..
} => handle_head(garage, &req, bucket_id, &key, part_number).await,
Endpoint::GetObject {
key, part_number, ..
} => handle_get(garage, &req, bucket_id, &key, part_number).await,
Endpoint::UploadPart {
key,
part_number,
upload_id,
} => {
handle_put_part(
garage,
req,
bucket_id,
&key,
part_number,
&upload_id,
content_sha256,
)
.await
}
Endpoint::CopyObject { key } => {
handle_copy(garage, &api_key, &req, bucket_id, &key).await
}
Endpoint::UploadPartCopy {
key,
part_number,
upload_id,
} => {
handle_upload_part_copy(
garage,
&api_key,
&req,
bucket_id,
&key,
part_number,
&upload_id,
)
.await
}
Endpoint::PutObject { key } => {
handle_put(garage, req, &bucket, &key, content_sha256).await
}
Endpoint::AbortMultipartUpload { key, upload_id } => {
handle_abort_multipart_upload(garage, bucket_id, &key, &upload_id).await
}
Endpoint::DeleteObject { key, .. } => handle_delete(garage, bucket_id, &key).await,
Endpoint::CreateMultipartUpload { key } => {
handle_create_multipart_upload(garage, &req, &bucket_name, bucket_id, &key).await
}
Endpoint::CompleteMultipartUpload { key, upload_id } => {
handle_complete_multipart_upload(
garage,
req,
&bucket_name,
&bucket,
&key,
&upload_id,
content_sha256,
)
.await
}
Endpoint::CreateBucket {} => unreachable!(),
Endpoint::HeadBucket {} => {
let empty_body: Body = Body::from(vec![]);
let response = Response::builder().body(empty_body).unwrap();
Ok(response)
}
Endpoint::DeleteBucket {} => {
handle_delete_bucket(&garage, bucket_id, bucket_name, api_key).await
}
Endpoint::GetBucketLocation {} => handle_get_bucket_location(garage),
Endpoint::GetBucketVersioning {} => handle_get_bucket_versioning(),
Endpoint::ListObjects {
delimiter,
encoding_type,
marker,
max_keys,
prefix,
} => {
handle_list(
garage,
&ListObjectsQuery {
common: ListQueryCommon {
bucket_name,
bucket_id,
delimiter: delimiter.map(|d| d.to_string()),
page_size: max_keys.map(|p| p.clamp(1, 1000)).unwrap_or(1000),
prefix: prefix.unwrap_or_default(),
urlencode_resp: encoding_type.map(|e| e == "url").unwrap_or(false),
},
is_v2: false,
marker,
continuation_token: None,
start_after: None,
},
)
.await
}
Endpoint::ListObjectsV2 {
delimiter,
encoding_type,
max_keys,
prefix,
continuation_token,
start_after,
list_type,
..
} => {
if list_type == "2" {
handle_list(
garage,
&ListObjectsQuery {
common: ListQueryCommon {
bucket_name,
bucket_id,
delimiter: delimiter.map(|d| d.to_string()),
page_size: max_keys.map(|p| p.clamp(1, 1000)).unwrap_or(1000),
urlencode_resp: encoding_type.map(|e| e == "url").unwrap_or(false),
prefix: prefix.unwrap_or_default(),
},
is_v2: true,
marker: None,
continuation_token,
start_after,
},
)
.await
} else {
Err(Error::bad_request(format!(
"Invalid endpoint: list-type={}",
list_type
)))
}
}
Endpoint::ListMultipartUploads {
delimiter,
encoding_type,
key_marker,
max_uploads,
prefix,
upload_id_marker,
} => {
handle_list_multipart_upload(
garage,
&ListMultipartUploadsQuery {
common: ListQueryCommon {
bucket_name,
bucket_id,
delimiter: delimiter.map(|d| d.to_string()),
page_size: max_uploads.map(|p| p.clamp(1, 1000)).unwrap_or(1000),
prefix: prefix.unwrap_or_default(),
urlencode_resp: encoding_type.map(|e| e == "url").unwrap_or(false),
},
key_marker,
upload_id_marker,
},
)
.await
}
Endpoint::ListParts {
key,
max_parts,
part_number_marker,
upload_id,
} => {
handle_list_parts(
garage,
&ListPartsQuery {
bucket_name,
bucket_id,
key,
upload_id,
part_number_marker: part_number_marker.map(|p| p.clamp(1, 10000)),
max_parts: max_parts.map(|p| p.clamp(1, 1000)).unwrap_or(1000),
},
)
.await
}
Endpoint::DeleteObjects {} => {
handle_delete_objects(garage, bucket_id, req, content_sha256).await
}
Endpoint::GetBucketWebsite {} => handle_get_website(&bucket).await,
Endpoint::PutBucketWebsite {} => {
handle_put_website(garage, bucket_id, req, content_sha256).await
}
Endpoint::DeleteBucketWebsite {} => handle_delete_website(garage, bucket_id).await,
Endpoint::GetBucketCors {} => handle_get_cors(&bucket).await,
Endpoint::PutBucketCors {} => {
handle_put_cors(garage, bucket_id, req, content_sha256).await
}
Endpoint::DeleteBucketCors {} => handle_delete_cors(garage, bucket_id).await,
endpoint => Err(Error::NotImplemented(endpoint.name().to_owned())),
};
// If request was a success and we have a CORS rule that applies to it,
// add the corresponding CORS headers to the response
let mut resp_ok = resp?;
if let Some(rule) = matching_cors_rule {
add_cors_headers(&mut resp_ok, rule)
.ok_or_internal_error("Invalid bucket CORS configuration")?;
}
Ok(resp_ok)
}
}
impl ApiEndpoint for S3ApiEndpoint {
fn name(&self) -> &'static str {
self.endpoint.name()
}
fn add_span_attributes(&self, span: SpanRef<'_>) {
span.set_attribute(KeyValue::new(
"bucket",
self.bucket_name.clone().unwrap_or_default(),
));
}
}
+8 -13
View File
@@ -7,15 +7,15 @@ use garage_model::bucket_alias_table::*;
use garage_model::bucket_table::Bucket;
use garage_model::garage::Garage;
use garage_model::key_table::Key;
use garage_model::object_table::ObjectFilter;
use garage_model::permission::BucketKeyPerm;
use garage_table::util::*;
use garage_util::crdt::*;
use garage_util::data::*;
use garage_util::time::*;
use crate::error::*;
use crate::s3_xml;
use crate::common_error::CommonError;
use crate::s3::error::*;
use crate::s3::xml as s3_xml;
use crate::signature::verify_signed_content;
pub fn handle_get_bucket_location(garage: Arc<Garage>) -> Result<Response<Body>, Error> {
@@ -130,7 +130,7 @@ pub async fn handle_create_bucket(
if let Some(location_constraint) = cmd {
if location_constraint != garage.config.s3_api.s3_region {
return Err(Error::BadRequest(format!(
return Err(Error::bad_request(format!(
"Cannot satisfy location constraint `{}`: buckets can only be created in region `{}`",
location_constraint,
garage.config.s3_api.s3_region
@@ -158,12 +158,12 @@ pub async fn handle_create_bucket(
// otherwise return a forbidden error.
let kp = api_key.bucket_permissions(&bucket_id);
if !(kp.allow_write || kp.allow_owner) {
return Err(Error::BucketAlreadyExists);
return Err(CommonError::BucketAlreadyExists.into());
}
} else {
// Create the bucket!
if !is_valid_bucket_name(&bucket_name) {
return Err(Error::BadRequest(format!(
return Err(Error::bad_request(format!(
"{}: {}",
bucket_name, INVALID_BUCKET_NAME_MESSAGE
)));
@@ -228,12 +228,8 @@ pub async fn handle_delete_bucket(
// Delete bucket
// Check bucket is empty
let objects = garage
.object_table
.get_range(&bucket_id, None, Some(ObjectFilter::IsData), 10)
.await?;
if !objects.is_empty() {
return Err(Error::BucketNotEmpty);
if !garage.bucket_helper().is_bucket_empty(bucket_id).await? {
return Err(CommonError::BucketNotEmpty.into());
}
// --- done checking, now commit ---
@@ -299,7 +295,6 @@ fn parse_create_bucket_xml(xml_bytes: &[u8]) -> Option<Option<String>> {
let mut ret = None;
for item in cbc.children() {
println!("{:?}", item);
if item.has_tag_name("LocationConstraint") {
if ret != None {
return None;
+38 -28
View File
@@ -5,23 +5,26 @@ use std::time::{Duration, SystemTime, UNIX_EPOCH};
use futures::{stream, stream::Stream, StreamExt, TryFutureExt};
use md5::{Digest as Md5Digest, Md5};
use bytes::Bytes;
use hyper::{Body, Request, Response};
use serde::Serialize;
use garage_rpc::netapp::bytes_buf::BytesBuf;
use garage_rpc::rpc_helper::OrderTag;
use garage_table::*;
use garage_util::data::*;
use garage_util::time::*;
use garage_model::block_ref_table::*;
use garage_model::garage::Garage;
use garage_model::key_table::Key;
use garage_model::object_table::*;
use garage_model::version_table::*;
use garage_model::s3::block_ref_table::*;
use garage_model::s3::object_table::*;
use garage_model::s3::version_table::*;
use crate::api_server::{parse_bucket_key, resolve_bucket};
use crate::error::*;
use crate::s3_put::{decode_upload_id, get_headers};
use crate::s3_xml::{self, xmlns_tag};
use crate::helpers::parse_bucket_key;
use crate::s3::error::*;
use crate::s3::put::{decode_upload_id, get_headers};
use crate::s3::xml::{self as s3_xml, xmlns_tag};
pub async fn handle_copy(
garage: Arc<Garage>,
@@ -201,8 +204,8 @@ pub async fn handle_upload_part_copy(
let mut ranges = http_range::HttpRange::parse(range_str, source_version_meta.size)
.map_err(|e| (e, source_version_meta.size))?;
if ranges.len() != 1 {
return Err(Error::BadRequest(
"Invalid x-amz-copy-source-range header: exactly 1 range must be given".into(),
return Err(Error::bad_request(
"Invalid x-amz-copy-source-range header: exactly 1 range must be given",
));
} else {
ranges.pop().unwrap()
@@ -230,8 +233,8 @@ pub async fn handle_upload_part_copy(
// This is only for small files, we don't bother handling this.
// (in AWS UploadPartCopy works for parts at least 5MB which
// is never the case of an inline object)
return Err(Error::BadRequest(
"Source object is too small (minimum part size is 5Mb)".into(),
return Err(Error::bad_request(
"Source object is too small (minimum part size is 5Mb)",
));
}
ObjectVersionData::FirstBlock(_meta, _first_block_hash) => (),
@@ -250,7 +253,7 @@ pub async fn handle_upload_part_copy(
// Check this part number hasn't yet been uploaded
if let Some(dv) = dest_version {
if dv.has_part_number(part_number) {
return Err(Error::BadRequest(format!(
return Err(Error::bad_request(format!(
"Part number {} has already been uploaded",
part_number
)));
@@ -305,13 +308,18 @@ pub async fn handle_upload_part_copy(
// if and only if the block returned is a block that already existed
// in the Garage data store (thus we don't need to save it again).
let garage2 = garage.clone();
let order_stream = OrderTag::stream();
let source_blocks = stream::iter(blocks_to_copy)
.flat_map(|(block_hash, range_to_copy)| {
.enumerate()
.flat_map(|(i, (block_hash, range_to_copy))| {
let garage3 = garage2.clone();
stream::once(async move {
let data = garage3.block_manager.rpc_get_block(&block_hash).await?;
let data = garage3
.block_manager
.rpc_get_block(&block_hash, Some(order_stream.order(i as u64)))
.await?;
match range_to_copy {
Some(r) => Ok((data[r].to_vec(), None)),
Some(r) => Ok((data.slice(r), None)),
None => Ok((data, Some(block_hash))),
}
})
@@ -413,10 +421,13 @@ async fn get_copy_source(
let copy_source = percent_encoding::percent_decode_str(copy_source).decode_utf8()?;
let (source_bucket, source_key) = parse_bucket_key(&copy_source, None)?;
let source_bucket_id = resolve_bucket(garage, &source_bucket.to_string(), api_key).await?;
let source_bucket_id = garage
.bucket_helper()
.resolve_bucket(&source_bucket.to_string(), api_key)
.await?;
if !api_key.allow_read(&source_bucket_id) {
return Err(Error::Forbidden(format!(
return Err(Error::forbidden(format!(
"Reading from bucket {} not allowed for this key",
source_bucket
)));
@@ -536,8 +547,8 @@ impl CopyPreconditionHeaders {
(None, None, None, Some(ims)) => v_date > *ims,
(None, None, None, None) => true,
_ => {
return Err(Error::BadRequest(
"Invalid combination of x-amz-copy-source-if-xxxxx headers".into(),
return Err(Error::bad_request(
"Invalid combination of x-amz-copy-source-if-xxxxx headers",
))
}
};
@@ -550,13 +561,13 @@ impl CopyPreconditionHeaders {
}
}
type BlockStreamItemOk = (Vec<u8>, Option<Hash>);
type BlockStreamItemOk = (Bytes, Option<Hash>);
type BlockStreamItem = Result<BlockStreamItemOk, garage_util::error::Error>;
struct Defragmenter<S: Stream<Item = BlockStreamItem>> {
block_size: usize,
block_stream: Pin<Box<stream::Peekable<S>>>,
buffer: Vec<u8>,
buffer: BytesBuf,
hash: Option<Hash>,
}
@@ -565,7 +576,7 @@ impl<S: Stream<Item = BlockStreamItem>> Defragmenter<S> {
Self {
block_size,
block_stream,
buffer: vec![],
buffer: BytesBuf::new(),
hash: None,
}
}
@@ -583,7 +594,7 @@ impl<S: Stream<Item = BlockStreamItem>> Defragmenter<S> {
if self.buffer.is_empty() {
let (next_block, next_block_hash) = self.block_stream.next().await.unwrap()?;
self.buffer = next_block;
self.buffer.extend(next_block);
self.hash = next_block_hash;
} else if self.buffer.len() + peeked_next_block.len() > self.block_size {
break;
@@ -594,11 +605,11 @@ impl<S: Stream<Item = BlockStreamItem>> Defragmenter<S> {
}
}
Ok((std::mem::take(&mut self.buffer), self.hash.take()))
Ok((self.buffer.take_all(), self.hash.take()))
}
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct CopyObjectResult {
#[serde(rename = "LastModified")]
pub last_modified: s3_xml::Value,
@@ -606,7 +617,7 @@ pub struct CopyObjectResult {
pub etag: s3_xml::Value,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct CopyPartResult {
#[serde(serialize_with = "xmlns_tag")]
pub xmlns: (),
@@ -619,7 +630,7 @@ pub struct CopyPartResult {
#[cfg(test)]
mod tests {
use super::*;
use crate::s3_xml::to_xml_with_header;
use crate::s3::xml::to_xml_with_header;
#[test]
fn copy_object_result() -> Result<(), Error> {
@@ -651,7 +662,6 @@ mod tests {
last_modified: s3_xml::Value("2011-04-11T20:34:56.000Z".into()),
etag: s3_xml::Value("\"9b2cf535f27731c974343645a3985328\"".into()),
};
println!("{}", to_xml_with_header(&v)?);
assert_eq!(to_xml_with_header(&v)?, expected_retval);
+12 -24
View File
@@ -9,13 +9,12 @@ use hyper::{header::HeaderName, Body, Method, Request, Response, StatusCode};
use serde::{Deserialize, Serialize};
use crate::error::*;
use crate::s3_xml::{to_xml_with_header, xmlns_tag, IntValue, Value};
use crate::s3::error::*;
use crate::s3::xml::{to_xml_with_header, xmlns_tag, IntValue, Value};
use crate::signature::verify_signed_content;
use garage_model::bucket_table::{Bucket, CorsRule as GarageCorsRule};
use garage_model::garage::Garage;
use garage_table::*;
use garage_util::data::*;
pub async fn handle_get_cors(bucket: &Bucket) -> Result<Response<Body>, Error> {
@@ -48,14 +47,11 @@ pub async fn handle_delete_cors(
bucket_id: Uuid,
) -> Result<Response<Body>, Error> {
let mut bucket = garage
.bucket_table
.get(&EmptyKey, &bucket_id)
.await?
.ok_or(Error::NoSuchBucket)?;
.bucket_helper()
.get_existing_bucket(bucket_id)
.await?;
let param = bucket
.params_mut()
.ok_or_internal_error("Bucket should not be deleted at this point")?;
let param = bucket.params_mut().unwrap();
param.cors_config.update(None);
garage.bucket_table.insert(&bucket).await?;
@@ -78,14 +74,11 @@ pub async fn handle_put_cors(
}
let mut bucket = garage
.bucket_table
.get(&EmptyKey, &bucket_id)
.await?
.ok_or(Error::NoSuchBucket)?;
.bucket_helper()
.get_existing_bucket(bucket_id)
.await?;
let param = bucket
.params_mut()
.ok_or_internal_error("Bucket should not be deleted at this point")?;
let param = bucket.params_mut().unwrap();
let conf: CorsConfiguration = from_reader(&body as &[u8])?;
conf.validate()?;
@@ -119,12 +112,7 @@ pub async fn handle_options_s3api(
let helper = garage.bucket_helper();
let bucket_id = helper.resolve_global_bucket_name(&bn).await?;
if let Some(id) = bucket_id {
let bucket = garage
.bucket_table
.get(&EmptyKey, &id)
.await?
.filter(|b| !b.state.is_deleted())
.ok_or(Error::NoSuchBucket)?;
let bucket = garage.bucket_helper().get_existing_bucket(id).await?;
handle_options_for_bucket(req, &bucket)
} else {
// If there is a bucket name in the request, but that name
@@ -185,7 +173,7 @@ pub fn handle_options_for_bucket(
}
}
Err(Error::Forbidden("This CORS request is not allowed.".into()))
Err(Error::forbidden("This CORS request is not allowed."))
}
pub fn find_matching_cors_rule<'a>(
+10 -11
View File
@@ -6,10 +6,10 @@ use garage_util::data::*;
use garage_util::time::*;
use garage_model::garage::Garage;
use garage_model::object_table::*;
use garage_model::s3::object_table::*;
use crate::error::*;
use crate::s3_xml;
use crate::s3::error::*;
use crate::s3::xml as s3_xml;
use crate::signature::verify_signed_content;
async fn handle_delete_internal(
@@ -64,14 +64,13 @@ pub async fn handle_delete(
bucket_id: Uuid,
key: &str,
) -> Result<Response<Body>, Error> {
let (_deleted_version, delete_marker_version) =
handle_delete_internal(&garage, bucket_id, key).await?;
Ok(Response::builder()
.header("x-amz-version-id", hex::encode(delete_marker_version))
.status(StatusCode::NO_CONTENT)
.body(Body::from(vec![]))
.unwrap())
match handle_delete_internal(&garage, bucket_id, key).await {
Ok(_) | Err(Error::NoSuchKey) => Ok(Response::builder()
.status(StatusCode::NO_CONTENT)
.body(Body::from(vec![]))
.unwrap()),
Err(e) => Err(e),
}
}
pub async fn handle_delete_objects(
+89 -158
View File
@@ -2,34 +2,24 @@ use std::convert::TryInto;
use err_derive::Error;
use hyper::header::HeaderValue;
use hyper::{HeaderMap, StatusCode};
use hyper::{Body, HeaderMap, StatusCode};
use garage_model::helper::error::Error as HelperError;
use garage_util::error::Error as GarageError;
use crate::s3_xml;
use crate::common_error::CommonError;
pub use crate::common_error::{CommonErrorDerivative, OkOrBadRequest, OkOrInternalError};
use crate::generic_server::ApiError;
use crate::s3::xml as s3_xml;
use crate::signature::error::Error as SignatureError;
/// Errors of this crate
#[derive(Debug, Error)]
pub enum Error {
// Category: internal error
/// Error related to deeper parts of Garage
#[error(display = "Internal error: {}", _0)]
InternalError(#[error(source)] GarageError),
/// Error related to Hyper
#[error(display = "Internal error (Hyper error): {}", _0)]
Hyper(#[error(source)] hyper::Error),
/// Error related to HTTP
#[error(display = "Internal error (HTTP error): {}", _0)]
Http(#[error(source)] http::Error),
#[error(display = "{}", _0)]
/// Error from common error
Common(CommonError),
// Category: cannot process
/// No proper api key was used, or the signature was invalid
#[error(display = "Forbidden: {}", _0)]
Forbidden(String),
/// Authorization Header Malformed
#[error(display = "Authorization header malformed, expected scope: {}", _0)]
AuthorizationHeaderMalformed(String),
@@ -38,22 +28,10 @@ pub enum Error {
#[error(display = "Key not found")]
NoSuchKey,
/// The bucket requested don't exists
#[error(display = "Bucket not found")]
NoSuchBucket,
/// The multipart upload requested don't exists
#[error(display = "Upload not found")]
NoSuchUpload,
/// Tried to create a bucket that already exist
#[error(display = "Bucket already exists")]
BucketAlreadyExists,
/// Tried to delete a non-empty bucket
#[error(display = "Tried to delete a non-empty bucket")]
BucketNotEmpty,
/// Precondition failed (e.g. x-amz-copy-source-if-match)
#[error(display = "At least one of the preconditions you specified did not hold")]
PreconditionFailed,
@@ -80,10 +58,6 @@ pub enum Error {
#[error(display = "Invalid UTF-8: {}", _0)]
InvalidUtf8String(#[error(source)] std::string::FromUtf8Error),
/// Some base64 encoded data was badly encoded
#[error(display = "Invalid base64: {}", _0)]
InvalidBase64(#[error(source)] base64::DecodeError),
/// The client sent invalid XML data
#[error(display = "Invalid XML: {}", _0)]
InvalidXml(String),
@@ -96,15 +70,34 @@ pub enum Error {
#[error(display = "Invalid HTTP range: {:?}", _0)]
InvalidRange(#[error(from)] (http_range::HttpRangeParseError, u64)),
/// The client sent an invalid request
#[error(display = "Bad request: {}", _0)]
BadRequest(String),
/// The client sent a request for an action not supported by garage
#[error(display = "Unimplemented action: {}", _0)]
NotImplemented(String),
}
impl<T> From<T> for Error
where
CommonError: From<T>,
{
fn from(err: T) -> Self {
Error::Common(CommonError::from(err))
}
}
impl CommonErrorDerivative for Error {}
impl From<HelperError> for Error {
fn from(err: HelperError) -> Self {
match err {
HelperError::Internal(i) => Self::Common(CommonError::InternalError(i)),
HelperError::BadRequest(b) => Self::Common(CommonError::BadRequest(b)),
HelperError::InvalidBucketName(n) => Self::Common(CommonError::InvalidBucketName(n)),
HelperError::NoSuchBucket(n) => Self::Common(CommonError::NoSuchBucket(n)),
e => Self::bad_request(format!("{}", e)),
}
}
}
impl From<roxmltree::Error> for Error {
fn from(err: roxmltree::Error) -> Self {
Self::InvalidXml(format!("{}", err))
@@ -117,88 +110,71 @@ impl From<quick_xml::de::DeError> for Error {
}
}
impl From<HelperError> for Error {
fn from(err: HelperError) -> Self {
impl From<SignatureError> for Error {
fn from(err: SignatureError) -> Self {
match err {
HelperError::Internal(i) => Self::InternalError(i),
HelperError::BadRequest(b) => Self::BadRequest(b),
SignatureError::Common(c) => Self::Common(c),
SignatureError::AuthorizationHeaderMalformed(c) => {
Self::AuthorizationHeaderMalformed(c)
}
SignatureError::InvalidUtf8Str(i) => Self::InvalidUtf8Str(i),
SignatureError::InvalidHeader(h) => Self::InvalidHeader(h),
}
}
}
impl From<multer::Error> for Error {
fn from(err: multer::Error) -> Self {
Self::BadRequest(err.to_string())
Self::bad_request(err)
}
}
impl Error {
/// Get the HTTP status code that best represents the meaning of the error for the client
pub fn http_status_code(&self) -> StatusCode {
match self {
Error::NoSuchKey | Error::NoSuchBucket | Error::NoSuchUpload => StatusCode::NOT_FOUND,
Error::BucketNotEmpty | Error::BucketAlreadyExists => StatusCode::CONFLICT,
Error::PreconditionFailed => StatusCode::PRECONDITION_FAILED,
Error::Forbidden(_) => StatusCode::FORBIDDEN,
Error::InternalError(
GarageError::Timeout
| GarageError::RemoteError(_)
| GarageError::Quorum(_, _, _, _),
) => StatusCode::SERVICE_UNAVAILABLE,
Error::InternalError(_) | Error::Hyper(_) | Error::Http(_) => {
StatusCode::INTERNAL_SERVER_ERROR
}
Error::InvalidRange(_) => StatusCode::RANGE_NOT_SATISFIABLE,
Error::NotImplemented(_) => StatusCode::NOT_IMPLEMENTED,
_ => StatusCode::BAD_REQUEST,
}
}
pub fn aws_code(&self) -> &'static str {
match self {
Error::Common(c) => c.aws_code(),
Error::NoSuchKey => "NoSuchKey",
Error::NoSuchBucket => "NoSuchBucket",
Error::NoSuchUpload => "NoSuchUpload",
Error::BucketAlreadyExists => "BucketAlreadyExists",
Error::BucketNotEmpty => "BucketNotEmpty",
Error::PreconditionFailed => "PreconditionFailed",
Error::InvalidPart => "InvalidPart",
Error::InvalidPartOrder => "InvalidPartOrder",
Error::EntityTooSmall => "EntityTooSmall",
Error::Forbidden(_) => "AccessDenied",
Error::AuthorizationHeaderMalformed(_) => "AuthorizationHeaderMalformed",
Error::NotImplemented(_) => "NotImplemented",
Error::InternalError(
GarageError::Timeout
| GarageError::RemoteError(_)
| GarageError::Quorum(_, _, _, _),
) => "ServiceUnavailable",
Error::InternalError(_) | Error::Hyper(_) | Error::Http(_) => "InternalError",
_ => "InvalidRequest",
Error::InvalidXml(_) => "MalformedXML",
Error::InvalidRange(_) => "InvalidRange",
Error::InvalidUtf8Str(_) | Error::InvalidUtf8String(_) | Error::InvalidHeader(_) => {
"InvalidRequest"
}
}
}
}
impl ApiError for Error {
/// Get the HTTP status code that best represents the meaning of the error for the client
fn http_status_code(&self) -> StatusCode {
match self {
Error::Common(c) => c.http_status_code(),
Error::NoSuchKey | Error::NoSuchUpload => StatusCode::NOT_FOUND,
Error::PreconditionFailed => StatusCode::PRECONDITION_FAILED,
Error::InvalidRange(_) => StatusCode::RANGE_NOT_SATISFIABLE,
Error::NotImplemented(_) => StatusCode::NOT_IMPLEMENTED,
Error::AuthorizationHeaderMalformed(_)
| Error::InvalidPart
| Error::InvalidPartOrder
| Error::EntityTooSmall
| Error::InvalidXml(_)
| Error::InvalidUtf8Str(_)
| Error::InvalidUtf8String(_)
| Error::InvalidHeader(_) => StatusCode::BAD_REQUEST,
}
}
pub fn aws_xml(&self, garage_region: &str, path: &str) -> String {
let error = s3_xml::Error {
code: s3_xml::Value(self.aws_code().to_string()),
message: s3_xml::Value(format!("{}", self)),
resource: Some(s3_xml::Value(path.to_string())),
region: Some(s3_xml::Value(garage_region.to_string())),
};
s3_xml::to_xml_with_header(&error).unwrap_or_else(|_| {
r#"
<?xml version="1.0" encoding="UTF-8"?>
<Error>
<Code>InternalError</Code>
<Message>XML encoding of error failed</Message>
</Error>
"#
.into()
})
}
pub fn add_headers(&self, header_map: &mut HeaderMap<HeaderValue>) {
fn add_http_headers(&self, header_map: &mut HeaderMap<HeaderValue>) {
use hyper::header;
header_map.append(header::CONTENT_TYPE, "application/xml".parse().unwrap());
#[allow(clippy::single_match)]
match self {
Error::InvalidRange((_, len)) => {
@@ -212,68 +188,23 @@ impl Error {
_ => (),
}
}
}
/// Trait to map error to the Bad Request error code
pub trait OkOrBadRequest {
type S;
fn ok_or_bad_request<M: AsRef<str>>(self, reason: M) -> Result<Self::S, Error>;
}
impl<T, E> OkOrBadRequest for Result<T, E>
where
E: std::fmt::Display,
{
type S = T;
fn ok_or_bad_request<M: AsRef<str>>(self, reason: M) -> Result<T, Error> {
match self {
Ok(x) => Ok(x),
Err(e) => Err(Error::BadRequest(format!("{}: {}", reason.as_ref(), e))),
}
}
}
impl<T> OkOrBadRequest for Option<T> {
type S = T;
fn ok_or_bad_request<M: AsRef<str>>(self, reason: M) -> Result<T, Error> {
match self {
Some(x) => Ok(x),
None => Err(Error::BadRequest(reason.as_ref().to_string())),
}
}
}
/// Trait to map an error to an Internal Error code
pub trait OkOrInternalError {
type S;
fn ok_or_internal_error<M: AsRef<str>>(self, reason: M) -> Result<Self::S, Error>;
}
impl<T, E> OkOrInternalError for Result<T, E>
where
E: std::fmt::Display,
{
type S = T;
fn ok_or_internal_error<M: AsRef<str>>(self, reason: M) -> Result<T, Error> {
match self {
Ok(x) => Ok(x),
Err(e) => Err(Error::InternalError(GarageError::Message(format!(
"{}: {}",
reason.as_ref(),
e
)))),
}
}
}
impl<T> OkOrInternalError for Option<T> {
type S = T;
fn ok_or_internal_error<M: AsRef<str>>(self, reason: M) -> Result<T, Error> {
match self {
Some(x) => Ok(x),
None => Err(Error::InternalError(GarageError::Message(
reason.as_ref().to_string(),
))),
}
fn http_body(&self, garage_region: &str, path: &str) -> Body {
let error = s3_xml::Error {
code: s3_xml::Value(self.aws_code().to_string()),
message: s3_xml::Value(format!("{}", self)),
resource: Some(s3_xml::Value(path.to_string())),
region: Some(s3_xml::Value(garage_region.to_string())),
};
Body::from(s3_xml::to_xml_with_header(&error).unwrap_or_else(|_| {
r#"
<?xml version="1.0" encoding="UTF-8"?>
<Error>
<Code>InternalError</Code>
<Message>XML encoding of error failed</Message>
</Error>
"#
.into()
}))
}
}
+119 -57
View File
@@ -2,22 +2,25 @@
use std::sync::Arc;
use std::time::{Duration, UNIX_EPOCH};
use futures::stream::*;
use futures::future;
use futures::stream::{self, StreamExt};
use http::header::{
ACCEPT_RANGES, CONTENT_LENGTH, CONTENT_RANGE, CONTENT_TYPE, ETAG, IF_MODIFIED_SINCE,
IF_NONE_MATCH, LAST_MODIFIED, RANGE,
};
use hyper::body::Bytes;
use hyper::{Body, Request, Response, StatusCode};
use tokio::sync::mpsc;
use garage_rpc::rpc_helper::{netapp::stream::ByteStream, OrderTag};
use garage_table::EmptyKey;
use garage_util::data::*;
use garage_util::error::OkOrMessage;
use garage_model::garage::Garage;
use garage_model::object_table::*;
use garage_model::version_table::*;
use garage_model::s3::object_table::*;
use garage_model::s3::version_table::*;
use crate::error::*;
use crate::s3::error::*;
const X_AMZ_MP_PARTS_COUNT: &str = "x-amz-mp-parts-count";
@@ -210,8 +213,8 @@ pub async fn handle_get(
match (part_number, parse_range_header(req, last_v_meta.size)?) {
(Some(_), Some(_)) => {
return Err(Error::BadRequest(
"Cannot specify both partNumber and Range header".into(),
return Err(Error::bad_request(
"Cannot specify both partNumber and Range header",
));
}
(Some(pn), None) => {
@@ -242,36 +245,56 @@ pub async fn handle_get(
Ok(resp_builder.body(body)?)
}
ObjectVersionData::FirstBlock(_, first_block_hash) => {
let read_first_block = garage.block_manager.rpc_get_block(first_block_hash);
let get_next_blocks = garage.version_table.get(&last_v.uuid, &EmptyKey);
let (tx, rx) = mpsc::channel(2);
let (first_block, version) = futures::try_join!(read_first_block, get_next_blocks)?;
let version = version.ok_or(Error::NoSuchKey)?;
let order_stream = OrderTag::stream();
let first_block_hash = *first_block_hash;
let version_uuid = last_v.uuid;
let mut blocks = version
.blocks
.items()
.iter()
.map(|(_, vb)| (vb.hash, None))
.collect::<Vec<_>>();
blocks[0].1 = Some(first_block);
tokio::spawn(async move {
match async {
let garage2 = garage.clone();
let version_fut = tokio::spawn(async move {
garage2.version_table.get(&version_uuid, &EmptyKey).await
});
let body_stream = futures::stream::iter(blocks)
.map(move |(hash, data_opt)| {
let garage = garage.clone();
async move {
if let Some(data) = data_opt {
Ok(Bytes::from(data))
} else {
garage
.block_manager
.rpc_get_block(&hash)
.await
.map(Bytes::from)
}
let stream_block_0 = garage
.block_manager
.rpc_get_block_streaming(&first_block_hash, Some(order_stream.order(0)))
.await?;
tx.send(stream_block_0)
.await
.ok_or_message("channel closed")?;
let version = version_fut.await.unwrap()?.ok_or(Error::NoSuchKey)?;
for (i, (_, vb)) in version.blocks.items().iter().enumerate().skip(1) {
let stream_block_i = garage
.block_manager
.rpc_get_block_streaming(&vb.hash, Some(order_stream.order(i as u64)))
.await?;
tx.send(stream_block_i)
.await
.ok_or_message("channel closed")?;
}
})
.buffered(2);
Ok::<(), Error>(())
}
.await
{
Ok(()) => (),
Err(e) => {
let err = std::io::Error::new(
std::io::ErrorKind::Other,
format!("Error while getting object data: {}", e),
);
let _ = tx
.send(Box::pin(stream::once(future::ready(Err(err)))))
.await;
}
}
});
let body_stream = tokio_stream::wrappers::ReceiverStream::new(rx).flatten();
let body = hyper::body::Body::wrap_stream(body_stream);
Ok(resp_builder.body(body)?)
@@ -302,9 +325,9 @@ async fn handle_get_range(
let body: Body = Body::from(bytes[begin as usize..end as usize].to_vec());
Ok(resp_builder.body(body)?)
} else {
None.ok_or_internal_error(
Err(Error::internal_error(
"Requested range not present in inline bytes when it should have been",
)
))
}
}
ObjectVersionData::FirstBlock(_meta, _first_block_hash) => {
@@ -422,40 +445,79 @@ fn body_from_blocks_range(
all_blocks.len(),
4 + ((end - begin) / std::cmp::max(all_blocks[0].1.size as u64, 1024)) as usize,
));
let mut true_offset = 0;
let mut block_offset: u64 = 0;
for (_, b) in all_blocks.iter() {
if true_offset >= end {
if block_offset >= end {
break;
}
// Keep only blocks that have an intersection with the requested range
if true_offset < end && true_offset + b.size > begin {
blocks.push((*b, true_offset));
if block_offset < end && block_offset + b.size > begin {
blocks.push((*b, block_offset));
}
true_offset += b.size;
block_offset += b.size as u64;
}
let order_stream = OrderTag::stream();
let body_stream = futures::stream::iter(blocks)
.map(move |(block, true_offset)| {
.enumerate()
.map(move |(i, (block, block_offset))| {
let garage = garage.clone();
async move {
let data = garage.block_manager.rpc_get_block(&block.hash).await?;
let data = Bytes::from(data);
let start_in_block = if true_offset > begin {
0
} else {
begin - true_offset
};
let end_in_block = if true_offset + block.size < end {
block.size
} else {
end - true_offset
};
Result::<Bytes, Error>::Ok(
data.slice(start_in_block as usize..end_in_block as usize),
)
garage
.block_manager
.rpc_get_block_streaming(&block.hash, Some(order_stream.order(i as u64)))
.await
.unwrap_or_else(|e| error_stream(i, e))
.scan(block_offset, move |chunk_offset, chunk| {
let r = match chunk {
Ok(chunk_bytes) => {
let chunk_len = chunk_bytes.len() as u64;
let r = if *chunk_offset >= end {
// The current chunk is after the part we want to read.
// Returning None here will stop the scan, the rest of the
// stream will be ignored
None
} else if *chunk_offset + chunk_len <= begin {
// The current chunk is before the part we want to read.
// We return a None that will be removed by the filter_map
// below.
Some(None)
} else {
// The chunk has an intersection with the requested range
let start_in_chunk = if *chunk_offset > begin {
0
} else {
begin - *chunk_offset
};
let end_in_chunk = if *chunk_offset + chunk_len < end {
chunk_len
} else {
end - *chunk_offset
};
Some(Some(Ok(chunk_bytes
.slice(start_in_chunk as usize..end_in_chunk as usize))))
};
*chunk_offset += chunk_bytes.len() as u64;
r
}
Err(e) => Some(Some(Err(e))),
};
futures::future::ready(r)
})
.filter_map(futures::future::ready)
}
})
.buffered(2);
.buffered(2)
.flatten();
hyper::body::Body::wrap_stream(body_stream)
}
fn error_stream(i: usize, e: garage_util::error::Error) -> ByteStream {
Box::pin(futures::stream::once(async move {
Err(std::io::Error::new(
std::io::ErrorKind::Other,
format!("Could not get block {}: {}", i, e),
))
}))
}
+33 -73
View File
@@ -10,15 +10,16 @@ use garage_util::error::Error as GarageError;
use garage_util::time::*;
use garage_model::garage::Garage;
use garage_model::object_table::*;
use garage_model::version_table::Version;
use garage_model::s3::object_table::*;
use garage_model::s3::version_table::Version;
use garage_table::EmptyKey;
use garage_table::{EmptyKey, EnumerationOrder};
use crate::encoding::*;
use crate::error::*;
use crate::s3_put;
use crate::s3_xml;
use crate::helpers::key_after_prefix;
use crate::s3::error::*;
use crate::s3::put as s3_put;
use crate::s3::xml as s3_xml;
const DUMMY_NAME: &str = "Dummy Key";
const DUMMY_KEY: &str = "GKDummyKey";
@@ -66,8 +67,14 @@ pub async fn handle_list(
let io = |bucket, key, count| {
let t = &garage.object_table;
async move {
t.get_range(&bucket, key, Some(ObjectFilter::IsData), count)
.await
t.get_range(
&bucket,
key,
Some(ObjectFilter::IsData),
count,
EnumerationOrder::Forward,
)
.await
}
};
@@ -165,8 +172,14 @@ pub async fn handle_list_multipart_upload(
let io = |bucket, key, count| {
let t = &garage.object_table;
async move {
t.get_range(&bucket, key, Some(ObjectFilter::IsUploading), count)
.await
t.get_range(
&bucket,
key,
Some(ObjectFilter::IsUploading),
count,
EnumerationOrder::Forward,
)
.await
}
};
@@ -569,13 +582,19 @@ impl ListObjectsQuery {
// representing the key to start with.
(Some(token), _) => match &token[..1] {
"[" => Ok(RangeBegin::IncludingKey {
key: String::from_utf8(base64::decode(token[1..].as_bytes())?)?,
key: String::from_utf8(
base64::decode(token[1..].as_bytes())
.ok_or_bad_request("Invalid continuation token")?,
)?,
fallback_key: None,
}),
"]" => Ok(RangeBegin::AfterKey {
key: String::from_utf8(base64::decode(token[1..].as_bytes())?)?,
key: String::from_utf8(
base64::decode(token[1..].as_bytes())
.ok_or_bad_request("Invalid continuation token")?,
)?,
}),
_ => Err(Error::BadRequest("Invalid continuation token".to_string())),
_ => Err(Error::bad_request("Invalid continuation token")),
},
// StartAfter has defined semantics in the spec:
@@ -923,39 +942,13 @@ fn uriencode_maybe(s: &str, yes: bool) -> s3_xml::Value {
}
}
const UTF8_BEFORE_LAST_CHAR: char = '\u{10FFFE}';
/// Compute the key after the prefix
fn key_after_prefix(pfx: &str) -> Option<String> {
let mut next = pfx.to_string();
while !next.is_empty() {
let tail = next.pop().unwrap();
if tail >= char::MAX {
continue;
}
// Circumvent a limitation of RangeFrom that overflow earlier than needed
// See: https://doc.rust-lang.org/core/ops/struct.RangeFrom.html
let new_tail = if tail == UTF8_BEFORE_LAST_CHAR {
char::MAX
} else {
(tail..).nth(1).unwrap()
};
next.push(new_tail);
return Some(next);
}
None
}
/*
* Unit tests of this module
*/
#[cfg(test)]
mod tests {
use super::*;
use garage_model::version_table::*;
use garage_model::s3::version_table::*;
use garage_util::*;
use std::iter::FromIterator;
@@ -1002,39 +995,6 @@ mod tests {
}
}
#[test]
fn test_key_after_prefix() {
assert_eq!(UTF8_BEFORE_LAST_CHAR as u32, (char::MAX as u32) - 1);
assert_eq!(key_after_prefix("a/b/").unwrap().as_str(), "a/b0");
assert_eq!(key_after_prefix("").unwrap().as_str(), "");
assert_eq!(
key_after_prefix("􏿽").unwrap().as_str(),
String::from(char::from_u32(0x10FFFE).unwrap())
);
// When the last character is the biggest UTF8 char
let a = String::from_iter(['a', char::MAX].iter());
assert_eq!(key_after_prefix(a.as_str()).unwrap().as_str(), "b");
// When all characters are the biggest UTF8 char
let b = String::from_iter([char::MAX; 3].iter());
assert!(key_after_prefix(b.as_str()).is_none());
// Check utf8 surrogates
let c = String::from('\u{D7FF}');
assert_eq!(
key_after_prefix(c.as_str()).unwrap().as_str(),
String::from('\u{E000}')
);
// Check the character before the biggest one
let d = String::from('\u{10FFFE}');
assert_eq!(
key_after_prefix(d.as_str()).unwrap().as_str(),
String::from(char::MAX)
);
}
#[test]
fn test_common_prefixes() {
let mut query = query();
+15
View File
@@ -0,0 +1,15 @@
pub mod api_server;
pub mod error;
mod bucket;
mod copy;
pub mod cors;
mod delete;
pub mod get;
mod list;
mod post_object;
mod put;
mod website;
mod router;
pub mod xml;
@@ -14,16 +14,15 @@ use serde::Deserialize;
use garage_model::garage::Garage;
use crate::api_server::resolve_bucket;
use crate::error::*;
use crate::s3_put::{get_headers, save_stream};
use crate::s3_xml;
use crate::s3::error::*;
use crate::s3::put::{get_headers, save_stream};
use crate::s3::xml as s3_xml;
use crate::signature::payload::{parse_date, verify_v4};
pub async fn handle_post_object(
garage: Arc<Garage>,
req: Request<Body>,
bucket: String,
bucket_name: String,
) -> Result<Response<Body>, Error> {
let boundary = req
.headers()
@@ -48,9 +47,7 @@ pub async fn handle_post_object(
let field = if let Some(field) = multipart.next_field().await? {
field
} else {
return Err(Error::BadRequest(
"Request did not contain a file".to_owned(),
));
return Err(Error::bad_request("Request did not contain a file"));
};
let name: HeaderName = if let Some(Ok(name)) = field.name().map(TryInto::try_into) {
name
@@ -66,14 +63,14 @@ pub async fn handle_post_object(
"tag" => (/* tag need to be reencoded, but we don't support them yet anyway */),
"acl" => {
if params.insert("x-amz-acl", content).is_some() {
return Err(Error::BadRequest(
"Field 'acl' provided more than one time".to_string(),
return Err(Error::bad_request(
"Field 'acl' provided more than one time",
));
}
}
_ => {
if params.insert(&name, content).is_some() {
return Err(Error::BadRequest(format!(
return Err(Error::bad_request(format!(
"Field '{}' provided more than one time",
name
)));
@@ -90,9 +87,7 @@ pub async fn handle_post_object(
.to_str()?;
let credential = params
.get("x-amz-credential")
.ok_or_else(|| {
Error::Forbidden("Garage does not support anonymous access yet".to_string())
})?
.ok_or_else(|| Error::forbidden("Garage does not support anonymous access yet"))?
.to_str()?;
let policy = params
.get("policy")
@@ -119,17 +114,31 @@ pub async fn handle_post_object(
};
let date = parse_date(date)?;
let api_key = verify_v4(&garage, credential, &date, signature, policy.as_bytes()).await?;
let api_key = verify_v4(
&garage,
"s3",
credential,
&date,
signature,
policy.as_bytes(),
)
.await?;
let bucket_id = resolve_bucket(&garage, &bucket, &api_key).await?;
let bucket_id = garage
.bucket_helper()
.resolve_bucket(&bucket_name, &api_key)
.await?;
if !api_key.allow_write(&bucket_id) {
return Err(Error::Forbidden(
"Operation is not allowed for this key.".to_string(),
));
return Err(Error::forbidden("Operation is not allowed for this key."));
}
let decoded_policy = base64::decode(&policy)?;
let bucket = garage
.bucket_helper()
.get_existing_bucket(bucket_id)
.await?;
let decoded_policy = base64::decode(&policy).ok_or_bad_request("Invalid policy")?;
let decoded_policy: Policy =
serde_json::from_slice(&decoded_policy).ok_or_bad_request("Invalid policy")?;
@@ -137,9 +146,7 @@ pub async fn handle_post_object(
.ok_or_bad_request("Invalid expiration date")?
.into();
if Utc::now() - expiration > Duration::zero() {
return Err(Error::BadRequest(
"Expiration date is in the paste".to_string(),
));
return Err(Error::bad_request("Expiration date is in the paste"));
}
let mut conditions = decoded_policy.into_conditions()?;
@@ -151,7 +158,7 @@ pub async fn handle_post_object(
"policy" | "x-amz-signature" => (), // this is always accepted, as it's required to validate other fields
"content-type" => {
let conds = conditions.params.remove("content-type").ok_or_else(|| {
Error::BadRequest(format!("Key '{}' is not allowed in policy", param_key))
Error::bad_request(format!("Key '{}' is not allowed in policy", param_key))
})?;
for cond in conds {
let ok = match cond {
@@ -161,7 +168,7 @@ pub async fn handle_post_object(
}
};
if !ok {
return Err(Error::BadRequest(format!(
return Err(Error::bad_request(format!(
"Key '{}' has value not allowed in policy",
param_key
)));
@@ -170,7 +177,7 @@ pub async fn handle_post_object(
}
"key" => {
let conds = conditions.params.remove("key").ok_or_else(|| {
Error::BadRequest(format!("Key '{}' is not allowed in policy", param_key))
Error::bad_request(format!("Key '{}' is not allowed in policy", param_key))
})?;
for cond in conds {
let ok = match cond {
@@ -178,7 +185,7 @@ pub async fn handle_post_object(
Operation::StartsWith(s) => key.starts_with(&s),
};
if !ok {
return Err(Error::BadRequest(format!(
return Err(Error::bad_request(format!(
"Key '{}' has value not allowed in policy",
param_key
)));
@@ -193,7 +200,7 @@ pub async fn handle_post_object(
continue;
}
let conds = conditions.params.remove(&param_key).ok_or_else(|| {
Error::BadRequest(format!("Key '{}' is not allowed in policy", param_key))
Error::bad_request(format!("Key '{}' is not allowed in policy", param_key))
})?;
for cond in conds {
let ok = match cond {
@@ -201,7 +208,7 @@ pub async fn handle_post_object(
Operation::StartsWith(s) => value.to_str()?.starts_with(s.as_str()),
};
if !ok {
return Err(Error::BadRequest(format!(
return Err(Error::bad_request(format!(
"Key '{}' has value not allowed in policy",
param_key
)));
@@ -212,7 +219,7 @@ pub async fn handle_post_object(
}
if let Some((param_key, _)) = conditions.params.iter().next() {
return Err(Error::BadRequest(format!(
return Err(Error::bad_request(format!(
"Key '{}' is required in policy, but no value was provided",
param_key
)));
@@ -225,7 +232,7 @@ pub async fn handle_post_object(
garage,
headers,
StreamLimiter::new(stream, conditions.content_length),
bucket_id,
&bucket,
&key,
None,
None,
@@ -242,7 +249,7 @@ pub async fn handle_post_object(
{
target
.query_pairs_mut()
.append_pair("bucket", &bucket)
.append_pair("bucket", &bucket_name)
.append_pair("key", &key)
.append_pair("etag", &etag);
let target = target.to_string();
@@ -287,7 +294,7 @@ pub async fn handle_post_object(
let xml = s3_xml::PostObject {
xmlns: (),
location: s3_xml::Value(location),
bucket: s3_xml::Value(bucket),
bucket: s3_xml::Value(bucket_name),
key: s3_xml::Value(key),
etag: s3_xml::Value(etag),
};
@@ -318,7 +325,7 @@ impl Policy {
match condition {
PolicyCondition::Equal(map) => {
if map.len() != 1 {
return Err(Error::BadRequest("Invalid policy item".to_owned()));
return Err(Error::bad_request("Invalid policy item"));
}
let (mut k, v) = map.into_iter().next().expect("size was verified");
k.make_ascii_lowercase();
@@ -326,7 +333,7 @@ impl Policy {
}
PolicyCondition::OtherOp([cond, mut key, value]) => {
if key.remove(0) != '$' {
return Err(Error::BadRequest("Invalid policy item".to_owned()));
return Err(Error::bad_request("Invalid policy item"));
}
key.make_ascii_lowercase();
match cond.as_str() {
@@ -339,7 +346,7 @@ impl Policy {
.or_default()
.push(Operation::StartsWith(value));
}
_ => return Err(Error::BadRequest("Invalid policy item".to_owned())),
_ => return Err(Error::bad_request("Invalid policy item")),
}
}
PolicyCondition::SizeRange(key, min, max) => {
@@ -347,7 +354,7 @@ impl Policy {
length.0 = length.0.max(min);
length.1 = length.1.min(max);
} else {
return Err(Error::BadRequest("Invalid policy item".to_owned()));
return Err(Error::bad_request("Invalid policy item"));
}
}
}
@@ -412,15 +419,15 @@ where
self.read += bytes.len() as u64;
// optimization to fail early when we know before the end it's too long
if self.length.end() < &self.read {
return Poll::Ready(Some(Err(Error::BadRequest(
"File size does not match policy".to_owned(),
return Poll::Ready(Some(Err(Error::bad_request(
"File size does not match policy",
))));
}
}
Poll::Ready(None) => {
if !self.length.contains(&self.read) {
return Poll::Ready(Some(Err(Error::BadRequest(
"File size does not match policy".to_owned(),
return Poll::Ready(Some(Err(Error::bad_request(
"File size does not match policy",
))));
}
}
+156 -61
View File
@@ -1,4 +1,4 @@
use std::collections::{BTreeMap, BTreeSet, VecDeque};
use std::collections::{BTreeMap, BTreeSet, HashMap};
use std::sync::Arc;
use futures::prelude::*;
@@ -8,25 +8,34 @@ use hyper::{Request, Response};
use md5::{digest::generic_array::*, Digest as Md5Digest, Md5};
use sha2::Sha256;
use opentelemetry::{
trace::{FutureExt as OtelFutureExt, TraceContextExt, Tracer},
Context,
};
use garage_rpc::netapp::bytes_buf::BytesBuf;
use garage_table::*;
use garage_util::async_hash::*;
use garage_util::data::*;
use garage_util::error::Error as GarageError;
use garage_util::time::*;
use garage_block::manager::INLINE_THRESHOLD;
use garage_model::block_ref_table::*;
use garage_model::bucket_table::Bucket;
use garage_model::garage::Garage;
use garage_model::object_table::*;
use garage_model::version_table::*;
use garage_model::index_counter::CountedItem;
use garage_model::s3::block_ref_table::*;
use garage_model::s3::object_table::*;
use garage_model::s3::version_table::*;
use crate::error::*;
use crate::s3_xml;
use crate::s3::error::*;
use crate::s3::xml as s3_xml;
use crate::signature::verify_signed_content;
pub async fn handle_put(
garage: Arc<Garage>,
req: Request<Body>,
bucket_id: Uuid,
bucket: &Bucket,
key: &str,
content_sha256: Option<Hash>,
) -> Result<Response<Body>, Error> {
@@ -46,7 +55,7 @@ pub async fn handle_put(
garage,
headers,
body,
bucket_id,
bucket,
key,
content_md5,
content_sha256,
@@ -59,7 +68,7 @@ pub(crate) async fn save_stream<S: Stream<Item = Result<Bytes, Error>> + Unpin>(
garage: Arc<Garage>,
headers: ObjectVersionHeaders,
body: S,
bucket_id: Uuid,
bucket: &Bucket,
key: &str,
content_md5: Option<String>,
content_sha256: Option<FixedBytes32>,
@@ -80,6 +89,7 @@ pub(crate) async fn save_stream<S: Stream<Item = Result<Bytes, Error>> + Unpin>(
let data_md5sum_hex = hex::encode(data_md5sum);
let data_sha256sum = sha256sum(&first_block[..]);
let size = first_block.len() as u64;
ensure_checksum_matches(
data_md5sum.as_slice(),
@@ -88,20 +98,22 @@ pub(crate) async fn save_stream<S: Stream<Item = Result<Bytes, Error>> + Unpin>(
content_sha256,
)?;
check_quotas(&garage, bucket, key, size).await?;
let object_version = ObjectVersion {
uuid: version_uuid,
timestamp: version_timestamp,
state: ObjectVersionState::Complete(ObjectVersionData::Inline(
ObjectVersionMeta {
headers,
size: first_block.len() as u64,
size,
etag: data_md5sum_hex.clone(),
},
first_block,
first_block.to_vec(),
)),
};
let object = Object::new(bucket_id, key.into(), vec![object_version]);
let object = Object::new(bucket.id, key.into(), vec![object_version]);
garage.object_table.insert(&object).await?;
return Ok((version_uuid, data_md5sum_hex));
@@ -114,36 +126,42 @@ pub(crate) async fn save_stream<S: Stream<Item = Result<Bytes, Error>> + Unpin>(
timestamp: version_timestamp,
state: ObjectVersionState::Uploading(headers.clone()),
};
let object = Object::new(bucket_id, key.into(), vec![object_version.clone()]);
let object = Object::new(bucket.id, key.into(), vec![object_version.clone()]);
garage.object_table.insert(&object).await?;
// Initialize corresponding entry in version table
// Write this entry now, even with empty block list,
// to prevent block_ref entries from being deleted (they can be deleted
// if the reference a version that isn't found in the version table)
let version = Version::new(version_uuid, bucket_id, key.into(), false);
let version = Version::new(version_uuid, bucket.id, key.into(), false);
garage.version_table.insert(&version).await?;
// Transfer data and verify checksum
let first_block_hash = blake2sum(&first_block[..]);
let tx_result = read_and_put_blocks(
&garage,
&version,
1,
first_block,
first_block_hash,
&mut chunker,
)
.await
.and_then(|(total_size, data_md5sum, data_sha256sum)| {
let first_block_hash = async_blake2sum(first_block.clone()).await;
let tx_result = (|| async {
let (total_size, data_md5sum, data_sha256sum) = read_and_put_blocks(
&garage,
&version,
1,
first_block,
first_block_hash,
&mut chunker,
)
.await?;
ensure_checksum_matches(
data_md5sum.as_slice(),
data_sha256sum,
content_md5.as_deref(),
content_sha256,
)
.map(|()| (total_size, data_md5sum))
});
)?;
check_quotas(&garage, bucket, key, total_size).await?;
Ok((total_size, data_md5sum))
})()
.await;
// If something went wrong, clean up
let (total_size, md5sum_arr) = match tx_result {
@@ -151,7 +169,7 @@ pub(crate) async fn save_stream<S: Stream<Item = Result<Bytes, Error>> + Unpin>(
Err(e) => {
// Mark object as aborted, this will free the blocks further down
object_version.state = ObjectVersionState::Aborted;
let object = Object::new(bucket_id, key.into(), vec![object_version.clone()]);
let object = Object::new(bucket.id, key.into(), vec![object_version.clone()]);
garage.object_table.insert(&object).await?;
return Err(e);
}
@@ -167,7 +185,7 @@ pub(crate) async fn save_stream<S: Stream<Item = Result<Bytes, Error>> + Unpin>(
},
first_block_hash,
));
let object = Object::new(bucket_id, key.into(), vec![object_version]);
let object = Object::new(bucket.id, key.into(), vec![object_version]);
garage.object_table.insert(&object).await?;
Ok((version_uuid, md5sum_hex))
@@ -183,8 +201,8 @@ fn ensure_checksum_matches(
) -> Result<(), Error> {
if let Some(expected_sha256) = content_sha256 {
if expected_sha256 != data_sha256sum {
return Err(Error::BadRequest(
"Unable to validate x-amz-content-sha256".to_string(),
return Err(Error::bad_request(
"Unable to validate x-amz-content-sha256",
));
} else {
trace!("Successfully validated x-amz-content-sha256");
@@ -192,9 +210,7 @@ fn ensure_checksum_matches(
}
if let Some(expected_md5) = content_md5 {
if expected_md5.trim_matches('"') != base64::encode(data_md5sum) {
return Err(Error::BadRequest(
"Unable to validate content-md5".to_string(),
));
return Err(Error::bad_request("Unable to validate content-md5"));
} else {
trace!("Successfully validated content-md5");
}
@@ -202,18 +218,85 @@ fn ensure_checksum_matches(
Ok(())
}
/// Check that inserting this object with this size doesn't exceed bucket quotas
async fn check_quotas(
garage: &Arc<Garage>,
bucket: &Bucket,
key: &str,
size: u64,
) -> Result<(), Error> {
let quotas = bucket.state.as_option().unwrap().quotas.get();
if quotas.max_objects.is_none() && quotas.max_size.is_none() {
return Ok(());
};
let key = key.to_string();
let (prev_object, counters) = futures::try_join!(
garage.object_table.get(&bucket.id, &key),
garage.object_counter_table.table.get(&bucket.id, &EmptyKey),
)?;
let counters = counters
.map(|x| x.filtered_values(&garage.system.ring.borrow()))
.unwrap_or_default();
let (prev_cnt_obj, prev_cnt_size) = match prev_object {
Some(o) => {
let prev_cnt = o.counts().into_iter().collect::<HashMap<_, _>>();
(
prev_cnt.get(OBJECTS).cloned().unwrap_or_default(),
prev_cnt.get(BYTES).cloned().unwrap_or_default(),
)
}
None => (0, 0),
};
let cnt_obj_diff = 1 - prev_cnt_obj;
let cnt_size_diff = size as i64 - prev_cnt_size;
if let Some(mo) = quotas.max_objects {
let current_objects = counters.get(OBJECTS).cloned().unwrap_or_default();
if cnt_obj_diff > 0 && current_objects + cnt_obj_diff > mo as i64 {
return Err(Error::forbidden(format!(
"Object quota is reached, maximum objects for this bucket: {}",
mo
)));
}
}
if let Some(ms) = quotas.max_size {
let current_size = counters.get(BYTES).cloned().unwrap_or_default();
if cnt_size_diff > 0 && current_size + cnt_size_diff > ms as i64 {
return Err(Error::forbidden(format!(
"Bucket size quota is reached, maximum total size of objects for this bucket: {}. The bucket is already {} bytes, and this object would add {} bytes.",
ms, current_size, size
)));
}
}
Ok(())
}
async fn read_and_put_blocks<S: Stream<Item = Result<Bytes, Error>> + Unpin>(
garage: &Garage,
version: &Version,
part_number: u64,
first_block: Vec<u8>,
first_block: Bytes,
first_block_hash: Hash,
chunker: &mut StreamChunker<S>,
) -> Result<(u64, GenericArray<u8, typenum::U16>, Hash), Error> {
let mut md5hasher = Md5::new();
let mut sha256hasher = Sha256::new();
md5hasher.update(&first_block[..]);
sha256hasher.update(&first_block[..]);
let tracer = opentelemetry::global::tracer("garage");
let md5hasher = AsyncHasher::<Md5>::new();
let sha256hasher = AsyncHasher::<Sha256>::new();
futures::future::join(
md5hasher.update(first_block.clone()),
sha256hasher.update(first_block.clone()),
)
.with_context(Context::current_with_span(
tracer.start("Hash first block (md5, sha256)"),
))
.await;
let mut next_offset = first_block.len();
let mut put_curr_version_block = put_block_meta(
@@ -235,9 +318,15 @@ async fn read_and_put_blocks<S: Stream<Item = Result<Bytes, Error>> + Unpin>(
chunker.next(),
)?;
if let Some(block) = next_block {
md5hasher.update(&block[..]);
sha256hasher.update(&block[..]);
let block_hash = blake2sum(&block[..]);
let (_, _, block_hash) = futures::future::join3(
md5hasher.update(block.clone()),
sha256hasher.update(block.clone()),
async_blake2sum(block.clone()),
)
.with_context(Context::current_with_span(
tracer.start("Hash block (md5, sha256, blake2)"),
))
.await;
let block_len = block.len();
put_curr_version_block = put_block_meta(
garage,
@@ -255,9 +344,9 @@ async fn read_and_put_blocks<S: Stream<Item = Result<Bytes, Error>> + Unpin>(
}
let total_size = next_offset as u64;
let data_md5sum = md5hasher.finalize();
let data_md5sum = md5hasher.finalize().await;
let data_sha256sum = sha256hasher.finalize();
let data_sha256sum = sha256hasher.finalize().await;
let data_sha256sum = Hash::try_from(&data_sha256sum[..]).unwrap();
Ok((total_size, data_md5sum, data_sha256sum))
@@ -297,7 +386,7 @@ struct StreamChunker<S: Stream<Item = Result<Bytes, Error>>> {
stream: S,
read_all: bool,
block_size: usize,
buf: VecDeque<u8>,
buf: BytesBuf,
}
impl<S: Stream<Item = Result<Bytes, Error>> + Unpin> StreamChunker<S> {
@@ -306,11 +395,11 @@ impl<S: Stream<Item = Result<Bytes, Error>> + Unpin> StreamChunker<S> {
stream,
read_all: false,
block_size,
buf: VecDeque::with_capacity(2 * block_size),
buf: BytesBuf::new(),
}
}
async fn next(&mut self) -> Result<Option<Vec<u8>>, Error> {
async fn next(&mut self) -> Result<Option<Bytes>, Error> {
while !self.read_all && self.buf.len() < self.block_size {
if let Some(block) = self.stream.next().await {
let bytes = block?;
@@ -323,12 +412,8 @@ impl<S: Stream<Item = Result<Bytes, Error>> + Unpin> StreamChunker<S> {
if self.buf.is_empty() {
Ok(None)
} else if self.buf.len() <= self.block_size {
let block = self.buf.drain(..).collect::<Vec<u8>>();
Ok(Some(block))
} else {
let block = self.buf.drain(..self.block_size).collect::<Vec<u8>>();
Ok(Some(block))
Ok(Some(self.buf.take_max(self.block_size)))
}
}
}
@@ -428,7 +513,7 @@ pub async fn handle_put_part(
// Check part hasn't already been uploaded
if let Some(v) = version {
if v.has_part_number(part_number) {
return Err(Error::BadRequest(format!(
return Err(Error::bad_request(format!(
"Part number {} has already been uploaded",
part_number
)));
@@ -437,7 +522,9 @@ pub async fn handle_put_part(
// Copy block to store
let version = Version::new(version_uuid, bucket_id, key, false);
let first_block_hash = blake2sum(&first_block[..]);
let first_block_hash = async_blake2sum(first_block.clone()).await;
let (_, data_md5sum, data_sha256sum) = read_and_put_blocks(
&garage,
&version,
@@ -475,7 +562,7 @@ pub async fn handle_complete_multipart_upload(
garage: Arc<Garage>,
req: Request<Body>,
bucket_name: &str,
bucket_id: Uuid,
bucket: &Bucket,
key: &str,
upload_id: &str,
content_sha256: Option<Hash>,
@@ -499,7 +586,7 @@ pub async fn handle_complete_multipart_upload(
// Get object and version
let key = key.to_string();
let (object, version) = futures::try_join!(
garage.object_table.get(&bucket_id, &key),
garage.object_table.get(&bucket.id, &key),
garage.version_table.get(&version_uuid, &EmptyKey),
)?;
@@ -513,7 +600,7 @@ pub async fn handle_complete_multipart_upload(
let version = version.ok_or(Error::NoSuchKey)?;
if version.blocks.is_empty() {
return Err(Error::BadRequest("No data was uploaded".to_string()));
return Err(Error::bad_request("No data was uploaded"));
}
let headers = match object_version.state {
@@ -574,8 +661,8 @@ pub async fn handle_complete_multipart_upload(
.map(|x| x.part_number)
.eq(block_parts.into_iter());
if !same_parts {
return Err(Error::BadRequest(
"Part numbers in block list and part list do not match. This can happen if a part was partially uploaded. Please abort the multipart upload and try again.".into(),
return Err(Error::bad_request(
"Part numbers in block list and part list do not match. This can happen if a part was partially uploaded. Please abort the multipart upload and try again."
));
}
@@ -592,6 +679,14 @@ pub async fn handle_complete_multipart_upload(
// Calculate total size of final object
let total_size = version.blocks.items().iter().map(|x| x.1.size).sum();
if let Err(e) = check_quotas(&garage, bucket, &key, total_size).await {
object_version.state = ObjectVersionState::Aborted;
let final_object = Object::new(bucket.id, key.clone(), vec![object_version]);
garage.object_table.insert(&final_object).await?;
return Err(e);
}
// Write final object version
object_version.state = ObjectVersionState::Complete(ObjectVersionData::FirstBlock(
ObjectVersionMeta {
@@ -602,7 +697,7 @@ pub async fn handle_complete_multipart_upload(
version.blocks.items()[0].1.hash,
));
let final_object = Object::new(bucket_id, key.clone(), vec![object_version]);
let final_object = Object::new(bucket.id, key.clone(), vec![object_version]);
garage.object_table.insert(&final_object).await?;
// Send response saying ok we're done
+13 -212
View File
@@ -1,131 +1,13 @@
use crate::error::{Error, OkOrBadRequest};
use std::borrow::Cow;
use hyper::header::HeaderValue;
use hyper::{HeaderMap, Method, Request};
/// This macro is used to generate very repetitive match {} blocks in this module
/// It is _not_ made to be used anywhere else
macro_rules! s3_match {
(@match $enum:expr , [ $($endpoint:ident,)* ]) => {{
// usage: s3_match {@match my_enum, [ VariantWithField1, VariantWithField2 ..] }
// returns true if the variant was one of the listed variants, false otherwise.
use Endpoint::*;
match $enum {
$(
$endpoint { .. } => true,
)*
_ => false
}
}};
(@extract $enum:expr , $param:ident, [ $($endpoint:ident,)* ]) => {{
// usage: s3_match {@extract my_enum, field_name, [ VariantWithField1, VariantWithField2 ..] }
// returns Some(field_value), or None if the variant was not one of the listed variants.
use Endpoint::*;
match $enum {
$(
$endpoint {$param, ..} => Some($param),
)*
_ => None
}
}};
(@gen_parser ($keyword:expr, $key:expr, $query:expr, $header:expr),
key: [$($kw_k:ident $(if $required_k:ident)? $(header $header_k:expr)? => $api_k:ident $(($($conv_k:ident :: $param_k:ident),*))?,)*],
no_key: [$($kw_nk:ident $(if $required_nk:ident)? $(if_header $header_nk:expr)? => $api_nk:ident $(($($conv_nk:ident :: $param_nk:ident),*))?,)*]) => {{
// usage: s3_match {@gen_parser (keyword, key, query, header),
// key: [
// SOME_KEYWORD => VariantWithKey,
// ...
// ],
// no_key: [
// SOME_KEYWORD => VariantWithoutKey,
// ...
// ]
// }
// See in from_{method} for more detailed usage.
use Endpoint::*;
use keywords::*;
match ($keyword, !$key.is_empty()){
$(
($kw_k, true) if true $(&& $query.$required_k.is_some())? $(&& $header.contains_key($header_k))? => Ok($api_k {
key: $key,
$($(
$param_k: s3_match!(@@parse_param $query, $conv_k, $param_k),
)*)?
}),
)*
$(
($kw_nk, false) $(if $query.$required_nk.is_some())? $(if $header.contains($header_nk))? => Ok($api_nk {
$($(
$param_nk: s3_match!(@@parse_param $query, $conv_nk, $param_nk),
)*)?
}),
)*
(kw, _) => Err(Error::BadRequest(format!("Invalid endpoint: {}", kw)))
}
}};
use crate::helpers::Authorization;
use crate::router_macros::{generateQueryParameters, router_match};
use crate::s3::error::*;
(@@parse_param $query:expr, query_opt, $param:ident) => {{
// extract optional query parameter
$query.$param.take().map(|param| param.into_owned())
}};
(@@parse_param $query:expr, query, $param:ident) => {{
// extract mendatory query parameter
$query.$param.take().ok_or_bad_request("Missing argument for endpoint")?.into_owned()
}};
(@@parse_param $query:expr, opt_parse, $param:ident) => {{
// extract and parse optional query parameter
// missing parameter is file, however parse error is reported as an error
$query.$param
.take()
.map(|param| param.parse())
.transpose()
.map_err(|_| Error::BadRequest("Failed to parse query parameter".to_owned()))?
}};
(@@parse_param $query:expr, parse, $param:ident) => {{
// extract and parse mandatory query parameter
// both missing and un-parseable parameters are reported as errors
$query.$param.take().ok_or_bad_request("Missing argument for endpoint")?
.parse()
.map_err(|_| Error::BadRequest("Failed to parse query parameter".to_owned()))?
}};
(@func
$(#[$doc:meta])*
pub enum Endpoint {
$(
$(#[$outer:meta])*
$variant:ident $({
$($name:ident: $ty:ty,)*
})?,
)*
}) => {
$(#[$doc])*
pub enum Endpoint {
$(
$(#[$outer])*
$variant $({
$($name: $ty, )*
})?,
)*
}
impl Endpoint {
pub fn name(&self) -> &'static str {
match self {
$(Endpoint::$variant $({ $($name: _,)* .. })? => stringify!($variant),)*
}
}
}
};
(@if ($($cond:tt)+) then ($($then:tt)*) else ($($else:tt)*)) => {
$($then)*
};
(@if () then ($($then:tt)*) else ($($else:tt)*)) => {
$($else)*
};
}
s3_match! {@func
router_match! {@func
/// List of all S3 API endpoints.
///
@@ -460,7 +342,7 @@ impl Endpoint {
Method::POST => Self::from_post(key, &mut query)?,
Method::PUT => Self::from_put(key, &mut query, req.headers())?,
Method::DELETE => Self::from_delete(key, &mut query)?,
_ => return Err(Error::BadRequest("Unknown method".to_owned())),
_ => return Err(Error::bad_request("Unknown method")),
};
if let Some(message) = query.nonempty_message() {
@@ -471,7 +353,7 @@ impl Endpoint {
/// Determine which endpoint a request is for, knowing it is a GET.
fn from_get(key: String, query: &mut QueryParameters<'_>) -> Result<Self, Error> {
s3_match! {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), key, query, None),
key: [
@@ -528,7 +410,7 @@ impl Endpoint {
/// Determine which endpoint a request is for, knowing it is a HEAD.
fn from_head(key: String, query: &mut QueryParameters<'_>) -> Result<Self, Error> {
s3_match! {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), key, query, None),
key: [
@@ -542,7 +424,7 @@ impl Endpoint {
/// Determine which endpoint a request is for, knowing it is a POST.
fn from_post(key: String, query: &mut QueryParameters<'_>) -> Result<Self, Error> {
s3_match! {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), key, query, None),
key: [
@@ -564,7 +446,7 @@ impl Endpoint {
query: &mut QueryParameters<'_>,
headers: &HeaderMap<HeaderValue>,
) -> Result<Self, Error> {
s3_match! {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), key, query, headers),
key: [
@@ -606,7 +488,7 @@ impl Endpoint {
/// Determine which endpoint a request is for, knowing it is a DELETE.
fn from_delete(key: String, query: &mut QueryParameters<'_>) -> Result<Self, Error> {
s3_match! {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), key, query, None),
key: [
@@ -636,7 +518,7 @@ impl Endpoint {
/// Get the key the request target. Returns None for requests which don't use a key.
#[allow(dead_code)]
pub fn get_key(&self) -> Option<&str> {
s3_match! {
router_match! {
@extract
self,
key,
@@ -673,7 +555,7 @@ impl Endpoint {
if let Endpoint::ListBuckets = self {
return Authorization::None;
};
let readonly = s3_match! {
let readonly = router_match! {
@match
self,
[
@@ -717,7 +599,7 @@ impl Endpoint {
SelectObjectContent,
]
};
let owner = s3_match! {
let owner = router_match! {
@match
self,
[
@@ -740,87 +622,6 @@ impl Endpoint {
}
}
/// What kind of authorization is required to perform a given action
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Authorization {
/// No authorization is required
None,
/// Having Read permission on bucket
Read,
/// Having Write permission on bucket
Write,
/// Having Owner permission on bucket
Owner,
}
/// This macro is used to generate part of the code in this module. It must be called only one, and
/// is useless outside of this module.
macro_rules! generateQueryParameters {
( $($rest:expr => $name:ident),* ) => {
/// Struct containing all query parameters used in endpoints. Think of it as an HashMap,
/// but with keys statically known.
#[derive(Debug, Default)]
struct QueryParameters<'a> {
keyword: Option<Cow<'a, str>>,
$(
$name: Option<Cow<'a, str>>,
)*
}
impl<'a> QueryParameters<'a> {
/// Build this struct from the query part of an URI.
fn from_query(query: &'a str) -> Result<Self, Error> {
let mut res: Self = Default::default();
for (k, v) in url::form_urlencoded::parse(query.as_bytes()) {
let repeated = match k.as_ref() {
$(
$rest => if !v.is_empty() {
res.$name.replace(v).is_some()
} else {
false
},
)*
_ => {
if k.starts_with("response-") || k.starts_with("X-Amz-") {
false
} else if v.as_ref().is_empty() {
if res.keyword.replace(k).is_some() {
return Err(Error::BadRequest("Multiple keywords".to_owned()));
}
continue;
} else {
debug!("Received an unknown query parameter: '{}'", k);
false
}
}
};
if repeated {
return Err(Error::BadRequest(format!(
"Query parameter repeated: '{}'",
k
)));
}
}
Ok(res)
}
/// Get an error message in case not all parameters where used when extracting them to
/// build an Enpoint variant
fn nonempty_message(&self) -> Option<&str> {
if self.keyword.is_some() {
Some("Keyword not used")
} $(
else if self.$name.is_some() {
Some(concat!("'", $rest, "'"))
}
)* else {
None
}
}
}
}
}
// parameter name => struct field
generateQueryParameters! {
"continuation-token" => continuation_token,
+22 -29
View File
@@ -4,13 +4,12 @@ use std::sync::Arc;
use hyper::{Body, Request, Response, StatusCode};
use serde::{Deserialize, Serialize};
use crate::error::*;
use crate::s3_xml::{to_xml_with_header, xmlns_tag, IntValue, Value};
use crate::s3::error::*;
use crate::s3::xml::{to_xml_with_header, xmlns_tag, IntValue, Value};
use crate::signature::verify_signed_content;
use garage_model::bucket_table::*;
use garage_model::garage::Garage;
use garage_table::*;
use garage_util::data::*;
pub async fn handle_get_website(bucket: &Bucket) -> Result<Response<Body>, Error> {
@@ -47,14 +46,11 @@ pub async fn handle_delete_website(
bucket_id: Uuid,
) -> Result<Response<Body>, Error> {
let mut bucket = garage
.bucket_table
.get(&EmptyKey, &bucket_id)
.await?
.ok_or(Error::NoSuchBucket)?;
.bucket_helper()
.get_existing_bucket(bucket_id)
.await?;
let param = bucket
.params_mut()
.ok_or_internal_error("Bucket should not be deleted at this point")?;
let param = bucket.params_mut().unwrap();
param.website_config.update(None);
garage.bucket_table.insert(&bucket).await?;
@@ -77,14 +73,11 @@ pub async fn handle_put_website(
}
let mut bucket = garage
.bucket_table
.get(&EmptyKey, &bucket_id)
.await?
.ok_or(Error::NoSuchBucket)?;
.bucket_helper()
.get_existing_bucket(bucket_id)
.await?;
let param = bucket
.params_mut()
.ok_or_internal_error("Bucket should not be deleted at this point")?;
let param = bucket.params_mut().unwrap();
let conf: WebsiteConfiguration = from_reader(&body as &[u8])?;
conf.validate()?;
@@ -176,8 +169,8 @@ impl WebsiteConfiguration {
|| self.index_document.is_some()
|| self.routing_rules.is_some())
{
return Err(Error::BadRequest(
"Bad XML: can't have RedirectAllRequestsTo and other fields".to_owned(),
return Err(Error::bad_request(
"Bad XML: can't have RedirectAllRequestsTo and other fields",
));
}
if let Some(ref ed) = self.error_document {
@@ -222,8 +215,8 @@ impl WebsiteConfiguration {
impl Key {
pub fn validate(&self) -> Result<(), Error> {
if self.key.0.is_empty() {
Err(Error::BadRequest(
"Bad XML: error document specified but empty".to_owned(),
Err(Error::bad_request(
"Bad XML: error document specified but empty",
))
} else {
Ok(())
@@ -234,8 +227,8 @@ impl Key {
impl Suffix {
pub fn validate(&self) -> Result<(), Error> {
if self.suffix.0.is_empty() | self.suffix.0.contains('/') {
Err(Error::BadRequest(
"Bad XML: index document is empty or contains /".to_owned(),
Err(Error::bad_request(
"Bad XML: index document is empty or contains /",
))
} else {
Ok(())
@@ -247,7 +240,7 @@ impl Target {
pub fn validate(&self) -> Result<(), Error> {
if let Some(ref protocol) = self.protocol {
if protocol.0 != "http" && protocol.0 != "https" {
return Err(Error::BadRequest("Bad XML: invalid protocol".to_owned()));
return Err(Error::bad_request("Bad XML: invalid protocol"));
}
}
Ok(())
@@ -269,19 +262,19 @@ impl Redirect {
pub fn validate(&self, has_prefix: bool) -> Result<(), Error> {
if self.replace_prefix.is_some() {
if self.replace_full.is_some() {
return Err(Error::BadRequest(
"Bad XML: both ReplaceKeyPrefixWith and ReplaceKeyWith are set".to_owned(),
return Err(Error::bad_request(
"Bad XML: both ReplaceKeyPrefixWith and ReplaceKeyWith are set",
));
}
if !has_prefix {
return Err(Error::BadRequest(
"Bad XML: ReplaceKeyPrefixWith is set, but KeyPrefixEquals isn't".to_owned(),
return Err(Error::bad_request(
"Bad XML: ReplaceKeyPrefixWith is set, but KeyPrefixEquals isn't",
));
}
}
if let Some(ref protocol) = self.protocol {
if protocol.0 != "http" && protocol.0 != "https" {
return Err(Error::BadRequest("Bad XML: invalid protocol".to_owned()));
return Err(Error::bad_request("Bad XML: invalid protocol"));
}
}
// TODO there are probably more invalide cases, but which ones?
+22 -22
View File
@@ -1,7 +1,7 @@
use quick_xml::se::to_string;
use serde::{Deserialize, Serialize, Serializer};
use crate::Error as ApiError;
use crate::s3::error::Error as ApiError;
pub fn to_xml_with_header<T: Serialize>(x: &T) -> Result<String, ApiError> {
let mut xml = r#"<?xml version="1.0" encoding="UTF-8"?>"#.to_string();
@@ -25,7 +25,7 @@ impl From<&str> for Value {
#[derive(Debug, Serialize, Deserialize, PartialEq, Eq, PartialOrd, Ord)]
pub struct IntValue(#[serde(rename = "$value")] pub i64);
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct Bucket {
#[serde(rename = "CreationDate")]
pub creation_date: Value,
@@ -33,7 +33,7 @@ pub struct Bucket {
pub name: Value,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct Owner {
#[serde(rename = "DisplayName")]
pub display_name: Value,
@@ -41,13 +41,13 @@ pub struct Owner {
pub id: Value,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct BucketList {
#[serde(rename = "Bucket")]
pub entries: Vec<Bucket>,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct ListAllMyBucketsResult {
#[serde(rename = "Buckets")]
pub buckets: BucketList,
@@ -55,7 +55,7 @@ pub struct ListAllMyBucketsResult {
pub owner: Owner,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct LocationConstraint {
#[serde(serialize_with = "xmlns_tag")]
pub xmlns: (),
@@ -63,7 +63,7 @@ pub struct LocationConstraint {
pub region: String,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct Deleted {
#[serde(rename = "Key")]
pub key: Value,
@@ -73,7 +73,7 @@ pub struct Deleted {
pub delete_marker_version_id: Value,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct Error {
#[serde(rename = "Code")]
pub code: Value,
@@ -85,7 +85,7 @@ pub struct Error {
pub region: Option<Value>,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct DeleteError {
#[serde(rename = "Code")]
pub code: Value,
@@ -97,7 +97,7 @@ pub struct DeleteError {
pub version_id: Option<Value>,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct DeleteResult {
#[serde(serialize_with = "xmlns_tag")]
pub xmlns: (),
@@ -107,7 +107,7 @@ pub struct DeleteResult {
pub errors: Vec<DeleteError>,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct InitiateMultipartUploadResult {
#[serde(serialize_with = "xmlns_tag")]
pub xmlns: (),
@@ -119,7 +119,7 @@ pub struct InitiateMultipartUploadResult {
pub upload_id: Value,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct CompleteMultipartUploadResult {
#[serde(serialize_with = "xmlns_tag")]
pub xmlns: (),
@@ -133,7 +133,7 @@ pub struct CompleteMultipartUploadResult {
pub etag: Value,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct Initiator {
#[serde(rename = "DisplayName")]
pub display_name: Value,
@@ -141,7 +141,7 @@ pub struct Initiator {
pub id: Value,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct ListMultipartItem {
#[serde(rename = "Initiated")]
pub initiated: Value,
@@ -157,7 +157,7 @@ pub struct ListMultipartItem {
pub storage_class: Value,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct ListMultipartUploadsResult {
#[serde(serialize_with = "xmlns_tag")]
pub xmlns: (),
@@ -187,7 +187,7 @@ pub struct ListMultipartUploadsResult {
pub encoding_type: Option<Value>,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct PartItem {
#[serde(rename = "ETag")]
pub etag: Value,
@@ -199,7 +199,7 @@ pub struct PartItem {
pub size: IntValue,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct ListPartsResult {
#[serde(serialize_with = "xmlns_tag")]
pub xmlns: (),
@@ -227,7 +227,7 @@ pub struct ListPartsResult {
pub storage_class: Value,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct ListBucketItem {
#[serde(rename = "Key")]
pub key: Value,
@@ -241,13 +241,13 @@ pub struct ListBucketItem {
pub storage_class: Value,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct CommonPrefix {
#[serde(rename = "Prefix")]
pub prefix: Value,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct ListBucketResult {
#[serde(serialize_with = "xmlns_tag")]
pub xmlns: (),
@@ -281,7 +281,7 @@ pub struct ListBucketResult {
pub common_prefixes: Vec<CommonPrefix>,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct VersioningConfiguration {
#[serde(serialize_with = "xmlns_tag")]
pub xmlns: (),
@@ -289,7 +289,7 @@ pub struct VersioningConfiguration {
pub status: Option<Value>,
}
#[derive(Debug, Serialize, PartialEq)]
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct PostObject {
#[serde(serialize_with = "xmlns_tag")]
pub xmlns: (),
+36
View File
@@ -0,0 +1,36 @@
use err_derive::Error;
use crate::common_error::CommonError;
pub use crate::common_error::{CommonErrorDerivative, OkOrBadRequest, OkOrInternalError};
/// Errors of this crate
#[derive(Debug, Error)]
pub enum Error {
#[error(display = "{}", _0)]
/// Error from common error
Common(CommonError),
/// Authorization Header Malformed
#[error(display = "Authorization header malformed, expected scope: {}", _0)]
AuthorizationHeaderMalformed(String),
// Category: bad request
/// The request contained an invalid UTF-8 sequence in its path or in other parameters
#[error(display = "Invalid UTF-8: {}", _0)]
InvalidUtf8Str(#[error(source)] std::str::Utf8Error),
/// The client sent a header with invalid value
#[error(display = "Invalid header value: {}", _0)]
InvalidHeader(#[error(source)] hyper::header::ToStrError),
}
impl<T> From<T> for Error
where
CommonError: From<T>,
{
fn from(err: T) -> Self {
Error::Common(CommonError::from(err))
}
}
impl CommonErrorDerivative for Error {}
+18 -12
View File
@@ -1,14 +1,15 @@
use chrono::{DateTime, Utc};
use hmac::{Hmac, Mac, NewMac};
use hmac::{Hmac, Mac};
use sha2::Sha256;
use garage_util::data::{sha256sum, Hash};
use crate::error::*;
pub mod error;
pub mod payload;
pub mod streaming;
use error::*;
pub const SHORT_DATE: &str = "%Y%m%d";
pub const LONG_DATETIME: &str = "%Y%m%dT%H%M%SZ";
@@ -16,7 +17,7 @@ type HmacSha256 = Hmac<Sha256>;
pub fn verify_signed_content(expected_sha256: Hash, body: &[u8]) -> Result<(), Error> {
if expected_sha256 != sha256sum(body) {
return Err(Error::BadRequest(
return Err(Error::bad_request(
"Request content hash does not match signed hash".to_string(),
));
}
@@ -28,20 +29,25 @@ pub fn signing_hmac(
secret_key: &str,
region: &str,
service: &str,
) -> Result<HmacSha256, crypto_mac::InvalidKeyLength> {
) -> Result<HmacSha256, crypto_common::InvalidLength> {
let secret = String::from("AWS4") + secret_key;
let mut date_hmac = HmacSha256::new_varkey(secret.as_bytes())?;
let mut date_hmac = HmacSha256::new_from_slice(secret.as_bytes())?;
date_hmac.update(datetime.format(SHORT_DATE).to_string().as_bytes());
let mut region_hmac = HmacSha256::new_varkey(&date_hmac.finalize().into_bytes())?;
let mut region_hmac = HmacSha256::new_from_slice(&date_hmac.finalize().into_bytes())?;
region_hmac.update(region.as_bytes());
let mut service_hmac = HmacSha256::new_varkey(&region_hmac.finalize().into_bytes())?;
let mut service_hmac = HmacSha256::new_from_slice(&region_hmac.finalize().into_bytes())?;
service_hmac.update(service.as_bytes());
let mut signing_hmac = HmacSha256::new_varkey(&service_hmac.finalize().into_bytes())?;
let mut signing_hmac = HmacSha256::new_from_slice(&service_hmac.finalize().into_bytes())?;
signing_hmac.update(b"aws4_request");
let hmac = HmacSha256::new_varkey(&signing_hmac.finalize().into_bytes())?;
let hmac = HmacSha256::new_from_slice(&signing_hmac.finalize().into_bytes())?;
Ok(hmac)
}
pub fn compute_scope(datetime: &DateTime<Utc>, region: &str) -> String {
format!("{}/{}/s3/aws4_request", datetime.format(SHORT_DATE), region,)
pub fn compute_scope(datetime: &DateTime<Utc>, region: &str, service: &str) -> String {
format!(
"{}/{}/{}/aws4_request",
datetime.format(SHORT_DATE),
region,
service
)
}
+18 -17
View File
@@ -11,14 +11,15 @@ use garage_util::data::Hash;
use garage_model::garage::Garage;
use garage_model::key_table::*;
use super::signing_hmac;
use super::{LONG_DATETIME, SHORT_DATE};
use super::LONG_DATETIME;
use super::{compute_scope, signing_hmac};
use crate::encoding::uri_encode;
use crate::error::*;
use crate::signature::error::*;
pub async fn check_payload_signature(
garage: &Garage,
service: &str,
request: &Request<Body>,
) -> Result<(Option<Key>, Option<Hash>), Error> {
let mut headers = HashMap::new();
@@ -64,6 +65,7 @@ pub async fn check_payload_signature(
let key = verify_v4(
garage,
service,
&authorization.credential,
&authorization.date,
&authorization.signature,
@@ -103,7 +105,7 @@ fn parse_authorization(
let (auth_kind, rest) = authorization.split_at(first_space);
if auth_kind != "AWS4-HMAC-SHA256" {
return Err(Error::BadRequest("Unsupported authorization method".into()));
return Err(Error::bad_request("Unsupported authorization method"));
}
let mut auth_params = HashMap::new();
@@ -127,10 +129,11 @@ fn parse_authorization(
let date = headers
.get("x-amz-date")
.ok_or_bad_request("Missing X-Amz-Date field")
.map_err(Error::from)
.and_then(|d| parse_date(d))?;
if Utc::now() - date > Duration::hours(24) {
return Err(Error::BadRequest("Date is too old".to_string()));
return Err(Error::bad_request("Date is too old".to_string()));
}
let auth = Authorization {
@@ -154,7 +157,7 @@ fn parse_query_authorization(
headers: &HashMap<String, String>,
) -> Result<Authorization, Error> {
if algorithm != "AWS4-HMAC-SHA256" {
return Err(Error::BadRequest(
return Err(Error::bad_request(
"Unsupported authorization method".to_string(),
));
}
@@ -177,10 +180,10 @@ fn parse_query_authorization(
.get("x-amz-expires")
.ok_or_bad_request("X-Amz-Expires not found in query parameters")?
.parse()
.map_err(|_| Error::BadRequest("X-Amz-Expires is not a number".to_string()))?;
.map_err(|_| Error::bad_request("X-Amz-Expires is not a number".to_string()))?;
if duration > 7 * 24 * 3600 {
return Err(Error::BadRequest(
return Err(Error::bad_request(
"X-Amz-Exprires may not exceed a week".to_string(),
));
}
@@ -188,10 +191,11 @@ fn parse_query_authorization(
let date = headers
.get("x-amz-date")
.ok_or_bad_request("Missing X-Amz-Date field")
.map_err(Error::from)
.and_then(|d| parse_date(d))?;
if Utc::now() - date > Duration::seconds(duration) {
return Err(Error::BadRequest("Date is too old".to_string()));
return Err(Error::bad_request("Date is too old".to_string()));
}
Ok(Authorization {
@@ -281,6 +285,7 @@ pub fn parse_date(date: &str) -> Result<DateTime<Utc>, Error> {
pub async fn verify_v4(
garage: &Garage,
service: &str,
credential: &str,
date: &DateTime<Utc>,
signature: &str,
@@ -288,11 +293,7 @@ pub async fn verify_v4(
) -> Result<Key, Error> {
let (key_id, scope) = parse_credential(credential)?;
let scope_expected = format!(
"{}/{}/s3/aws4_request",
date.format(SHORT_DATE),
garage.config.s3_api.s3_region
);
let scope_expected = compute_scope(date, &garage.config.s3_api.s3_region, service);
if scope != scope_expected {
return Err(Error::AuthorizationHeaderMalformed(scope.to_string()));
}
@@ -302,20 +303,20 @@ pub async fn verify_v4(
.get(&EmptyKey, &key_id)
.await?
.filter(|k| !k.state.is_deleted())
.ok_or_else(|| Error::Forbidden(format!("No such key: {}", &key_id)))?;
.ok_or_else(|| Error::forbidden(format!("No such key: {}", &key_id)))?;
let key_p = key.params().unwrap();
let mut hmac = signing_hmac(
date,
&key_p.secret_key,
&garage.config.s3_api.s3_region,
"s3",
service,
)
.ok_or_internal_error("Unable to build signing HMAC")?;
hmac.update(payload);
let our_signature = hex::encode(hmac.finalize().into_bytes());
if signature != our_signature {
return Err(Error::Forbidden("Invalid signature".to_string()));
return Err(Error::forbidden("Invalid signature".to_string()));
}
Ok(key)
+59 -10
View File
@@ -1,18 +1,67 @@
use std::pin::Pin;
use chrono::{DateTime, Utc};
use chrono::{DateTime, NaiveDateTime, Utc};
use futures::prelude::*;
use futures::task;
use garage_model::key_table::Key;
use hmac::Mac;
use hyper::body::Bytes;
use hyper::{Body, Request};
use garage_util::data::Hash;
use hmac::Mac;
use super::sha256sum;
use super::HmacSha256;
use super::LONG_DATETIME;
use super::{compute_scope, sha256sum, HmacSha256, LONG_DATETIME};
use crate::error::*;
use crate::signature::error::*;
pub fn parse_streaming_body(
api_key: &Key,
req: Request<Body>,
content_sha256: &mut Option<Hash>,
region: &str,
service: &str,
) -> Result<Request<Body>, Error> {
match req.headers().get("x-amz-content-sha256") {
Some(header) if header == "STREAMING-AWS4-HMAC-SHA256-PAYLOAD" => {
let signature = content_sha256
.take()
.ok_or_bad_request("No signature provided")?;
let secret_key = &api_key
.state
.as_option()
.ok_or_internal_error("Deleted key state")?
.secret_key;
let date = req
.headers()
.get("x-amz-date")
.ok_or_bad_request("Missing X-Amz-Date field")?
.to_str()?;
let date: NaiveDateTime = NaiveDateTime::parse_from_str(date, LONG_DATETIME)
.ok_or_bad_request("Invalid date")?;
let date: DateTime<Utc> = DateTime::from_utc(date, Utc);
let scope = compute_scope(&date, region, service);
let signing_hmac = crate::signature::signing_hmac(&date, secret_key, region, service)
.ok_or_internal_error("Unable to build signing HMAC")?;
Ok(req.map(move |body| {
Body::wrap_stream(
SignedPayloadStream::new(
body.map_err(Error::from),
signing_hmac,
date,
&scope,
signature,
)
.map_err(Error::from),
)
}))
}
_ => Ok(req),
}
}
/// Result of `sha256("")`
const EMPTY_STRING_HEX_DIGEST: &str =
@@ -38,7 +87,7 @@ fn compute_streaming_payload_signature(
let mut hmac = signing_hmac.clone();
hmac.update(string_to_sign.as_bytes());
Hash::try_from(&hmac.finalize().into_bytes()).ok_or_internal_error("Invalid signature")
Ok(Hash::try_from(&hmac.finalize().into_bytes()).ok_or_internal_error("Invalid signature")?)
}
mod payload {
@@ -114,10 +163,10 @@ impl From<SignedPayloadStreamError> for Error {
match err {
SignedPayloadStreamError::Stream(e) => e,
SignedPayloadStreamError::InvalidSignature => {
Error::BadRequest("Invalid payload signature".into())
Error::bad_request("Invalid payload signature")
}
SignedPayloadStreamError::Message(e) => {
Error::BadRequest(format!("Chunk format error: {}", e))
Error::bad_request(format!("Chunk format error: {}", e))
}
}
}
@@ -295,7 +344,7 @@ mod tests {
.with_timezone(&Utc);
let secret_key = "test";
let region = "test";
let scope = crate::signature::compute_scope(&datetime, region);
let scope = crate::signature::compute_scope(&datetime, region, "s3");
let signing_hmac =
crate::signature::signing_hmac(&datetime, secret_key, region, "s3").unwrap();
+12 -6
View File
@@ -1,6 +1,6 @@
[package]
name = "garage_block"
version = "0.7.0"
version = "0.8.0"
authors = ["Alex Auvolat <alex@adnab.me>"]
edition = "2018"
license = "AGPL-3.0"
@@ -14,20 +14,22 @@ path = "lib.rs"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
garage_rpc = { version = "0.7.0", path = "../rpc" }
garage_util = { version = "0.7.0", path = "../util" }
garage_table = { version = "0.7.0", path = "../table" }
garage_db = { version = "0.8.0", path = "../db" }
garage_rpc = { version = "0.8.0", path = "../rpc" }
garage_util = { version = "0.8.0", path = "../util" }
garage_table = { version = "0.8.0", path = "../table" }
opentelemetry = "0.17"
arc-swap = "1.5"
async-trait = "0.1.7"
bytes = "1.0"
hex = "0.4"
tracing = "0.1.30"
rand = "0.8"
zstd = { version = "0.9", default-features = false }
sled = "0.34"
async-compression = { version = "0.3", features = ["tokio", "zstd"] }
zstd = { version = "0.9", default-features = false }
rmp-serde = "0.15"
serde = { version = "1.0", default-features = false, features = ["derive", "rc"] }
@@ -36,3 +38,7 @@ serde_bytes = "0.11"
futures = "0.3"
futures-util = "0.3"
tokio = { version = "1.0", default-features = false, features = ["rt", "rt-multi-thread", "io-util", "net", "time", "macros", "sync", "signal", "fs"] }
tokio-util = { version = "0.6", features = ["io"] }
[features]
system-libs = [ "zstd/pkg-config" ]
+36 -12
View File
@@ -1,16 +1,22 @@
use bytes::Bytes;
use serde::{Deserialize, Serialize};
use zstd::stream::{decode_all as zstd_decode, Encoder};
use garage_util::data::*;
use garage_util::error::*;
#[derive(Debug, Serialize, Deserialize, Copy, Clone)]
pub enum DataBlockHeader {
Plain,
Compressed,
}
/// A possibly compressed block of data
#[derive(Debug, Serialize, Deserialize)]
pub enum DataBlock {
/// Uncompressed data
Plain(#[serde(with = "serde_bytes")] Vec<u8>),
Plain(Bytes),
/// Data compressed with zstd
Compressed(#[serde(with = "serde_bytes")] Vec<u8>),
Compressed(Bytes),
}
impl DataBlock {
@@ -30,7 +36,7 @@ impl DataBlock {
/// Get the buffer, possibly decompressing it, and verify it's integrity.
/// For Plain block, data is compared to hash, for Compressed block, zstd checksumming system
/// is used instead.
pub fn verify_get(self, hash: Hash) -> Result<Vec<u8>, Error> {
pub fn verify_get(self, hash: Hash) -> Result<Bytes, Error> {
match self {
DataBlock::Plain(data) => {
if blake2sum(&data) == hash {
@@ -39,9 +45,9 @@ impl DataBlock {
Err(Error::CorruptData(hash))
}
}
DataBlock::Compressed(data) => {
zstd_decode(&data[..]).map_err(|_| Error::CorruptData(hash))
}
DataBlock::Compressed(data) => zstd_decode(&data[..])
.map_err(|_| Error::CorruptData(hash))
.map(Bytes::from),
}
}
@@ -61,13 +67,31 @@ impl DataBlock {
}
}
pub fn from_buffer(data: Vec<u8>, level: Option<i32>) -> DataBlock {
if let Some(level) = level {
if let Ok(data) = zstd_encode(&data[..], level) {
return DataBlock::Compressed(data);
pub async fn from_buffer(data: Bytes, level: Option<i32>) -> DataBlock {
tokio::task::spawn_blocking(move || {
if let Some(level) = level {
if let Ok(data) = zstd_encode(&data[..], level) {
return DataBlock::Compressed(data.into());
}
}
DataBlock::Plain(data)
})
.await
.unwrap()
}
pub fn into_parts(self) -> (DataBlockHeader, Bytes) {
match self {
DataBlock::Plain(data) => (DataBlockHeader::Plain, data),
DataBlock::Compressed(data) => (DataBlockHeader::Compressed, data),
}
}
pub fn from_parts(h: DataBlockHeader, bytes: Bytes) -> Self {
match h {
DataBlockHeader::Plain => DataBlock::Plain(bytes),
DataBlockHeader::Compressed => DataBlock::Compressed(bytes),
}
DataBlock::Plain(data)
}
}
+2
View File
@@ -2,6 +2,8 @@
extern crate tracing;
pub mod manager;
pub mod repair;
pub mod resync;
mod block;
mod metrics;
+312 -631
View File
File diff suppressed because it is too large Load Diff
+2 -2
View File
@@ -1,6 +1,6 @@
use opentelemetry::{global, metrics::*};
use garage_util::sled_counter::SledCountedTree;
use garage_db::counted_tree_hack::CountedTree;
/// TableMetrics reference all counter used for metrics
pub struct BlockManagerMetrics {
@@ -23,7 +23,7 @@ pub struct BlockManagerMetrics {
}
impl BlockManagerMetrics {
pub fn new(resync_queue: SledCountedTree, resync_errors: SledCountedTree) -> Self {
pub fn new(resync_queue: CountedTree, resync_errors: CountedTree) -> Self {
let meter = global::meter("garage_model/block");
Self {
_resync_queue_len: meter
+32 -17
View File
@@ -1,5 +1,7 @@
use std::convert::TryInto;
use garage_db as db;
use garage_util::data::*;
use garage_util::error::*;
use garage_util::time::*;
@@ -7,31 +9,41 @@ use garage_util::time::*;
use crate::manager::BLOCK_GC_DELAY;
pub struct BlockRc {
pub(crate) rc: sled::Tree,
pub(crate) rc: db::Tree,
}
impl BlockRc {
pub(crate) fn new(rc: sled::Tree) -> Self {
pub(crate) fn new(rc: db::Tree) -> Self {
Self { rc }
}
/// Increment the reference counter associated to a hash.
/// Returns true if the RC goes from zero to nonzero.
pub(crate) fn block_incref(&self, hash: &Hash) -> Result<bool, Error> {
let old_rc = self
.rc
.fetch_and_update(&hash, |old| RcEntry::parse_opt(old).increment().serialize())?;
let old_rc = RcEntry::parse_opt(old_rc);
pub(crate) fn block_incref(
&self,
tx: &mut db::Transaction,
hash: &Hash,
) -> db::TxOpResult<bool> {
let old_rc = RcEntry::parse_opt(tx.get(&self.rc, &hash)?);
match old_rc.increment().serialize() {
Some(x) => tx.insert(&self.rc, &hash, x)?,
None => unreachable!(),
};
Ok(old_rc.is_zero())
}
/// Decrement the reference counter associated to a hash.
/// Returns true if the RC is now zero.
pub(crate) fn block_decref(&self, hash: &Hash) -> Result<bool, Error> {
let new_rc = self
.rc
.update_and_fetch(&hash, |old| RcEntry::parse_opt(old).decrement().serialize())?;
let new_rc = RcEntry::parse_opt(new_rc);
pub(crate) fn block_decref(
&self,
tx: &mut db::Transaction,
hash: &Hash,
) -> db::TxOpResult<bool> {
let new_rc = RcEntry::parse_opt(tx.get(&self.rc, &hash)?).decrement();
match new_rc.serialize() {
Some(x) => tx.insert(&self.rc, &hash, x)?,
None => tx.remove(&self.rc, &hash)?,
};
Ok(matches!(new_rc, RcEntry::Deletable { .. }))
}
@@ -44,12 +56,15 @@ impl BlockRc {
/// deletion time has passed
pub(crate) fn clear_deleted_block_rc(&self, hash: &Hash) -> Result<(), Error> {
let now = now_msec();
self.rc.update_and_fetch(&hash, |rcval| {
let updated = match RcEntry::parse_opt(rcval) {
RcEntry::Deletable { at_time } if now > at_time => RcEntry::Absent,
v => v,
self.rc.db().transaction(|mut tx| {
let rcval = RcEntry::parse_opt(tx.get(&self.rc, &hash)?);
match rcval {
RcEntry::Deletable { at_time } if now > at_time => {
tx.remove(&self.rc, &hash)?;
}
_ => (),
};
updated.serialize()
tx.commit(())
})?;
Ok(())
}
+466
View File
@@ -0,0 +1,466 @@
use core::ops::Bound;
use std::path::PathBuf;
use std::sync::Arc;
use std::time::Duration;
use async_trait::async_trait;
use serde::{Deserialize, Serialize};
use tokio::fs;
use tokio::select;
use tokio::sync::mpsc;
use tokio::sync::watch;
use garage_util::background::*;
use garage_util::data::*;
use garage_util::error::*;
use garage_util::persister::Persister;
use garage_util::time::*;
use garage_util::tranquilizer::Tranquilizer;
use crate::manager::*;
// Full scrub every 30 days
const SCRUB_INTERVAL: Duration = Duration::from_secs(3600 * 24 * 30);
// Scrub tranquility is initially set to 4, but can be changed in the CLI
// and the updated version is persisted over Garage restarts
const INITIAL_SCRUB_TRANQUILITY: u32 = 4;
// ---- ---- ----
// FIRST KIND OF REPAIR: FINDING MISSING BLOCKS/USELESS BLOCKS
// This is a one-shot repair operation that can be launched,
// checks everything, and then exits.
// ---- ---- ----
pub struct RepairWorker {
manager: Arc<BlockManager>,
next_start: Option<Hash>,
block_iter: Option<BlockStoreIterator>,
}
impl RepairWorker {
pub fn new(manager: Arc<BlockManager>) -> Self {
Self {
manager,
next_start: None,
block_iter: None,
}
}
}
#[async_trait]
impl Worker for RepairWorker {
fn name(&self) -> String {
"Block repair worker".into()
}
fn info(&self) -> Option<String> {
match self.block_iter.as_ref() {
None => {
let idx_bytes = self
.next_start
.as_ref()
.map(|x| x.as_slice())
.unwrap_or(&[]);
let idx_bytes = if idx_bytes.len() > 4 {
&idx_bytes[..4]
} else {
idx_bytes
};
Some(format!("Phase 1: {}", hex::encode(idx_bytes)))
}
Some(bi) => Some(format!("Phase 2: {:.2}% done", bi.progress() * 100.)),
}
}
async fn work(&mut self, _must_exit: &mut watch::Receiver<bool>) -> Result<WorkerState, Error> {
match self.block_iter.as_mut() {
None => {
// Phase 1: Repair blocks from RC table.
// We have to do this complicated two-step process where we first read a bunch
// of hashes from the RC table, and then insert them in the to-resync queue,
// because of SQLite. Basically, as long as we have an iterator on a DB table,
// we can't do anything else on the DB. The naive approach (which we had previously)
// of just iterating on the RC table and inserting items one to one in the resync
// queue can't work here, it would just provoke a deadlock in the SQLite adapter code.
// This is mostly because the Rust bindings for SQLite assume a worst-case scenario
// where SQLite is not compiled in thread-safe mode, so we have to wrap everything
// in a mutex (see db/sqlite_adapter.rs and discussion in PR #322).
// TODO: maybe do this with tokio::task::spawn_blocking ?
let mut batch_of_hashes = vec![];
let start_bound = match self.next_start.as_ref() {
None => Bound::Unbounded,
Some(x) => Bound::Excluded(x.as_slice()),
};
for entry in self
.manager
.rc
.rc
.range::<&[u8], _>((start_bound, Bound::Unbounded))?
{
let (hash, _) = entry?;
let hash = Hash::try_from(&hash[..]).unwrap();
batch_of_hashes.push(hash);
if batch_of_hashes.len() >= 1000 {
break;
}
}
if batch_of_hashes.is_empty() {
// move on to phase 2
self.block_iter = Some(BlockStoreIterator::new(&self.manager));
return Ok(WorkerState::Busy);
}
for hash in batch_of_hashes.into_iter() {
self.manager
.resync
.put_to_resync(&hash, Duration::from_secs(0))?;
self.next_start = Some(hash)
}
Ok(WorkerState::Busy)
}
Some(bi) => {
// Phase 2: Repair blocks actually on disk
// Lists all blocks on disk and adds them to the resync queue.
// This allows us to find blocks we are storing but don't actually need,
// so that we can offload them if necessary and then delete them locally.
if let Some(hash) = bi.next().await? {
self.manager
.resync
.put_to_resync(&hash, Duration::from_secs(0))?;
Ok(WorkerState::Busy)
} else {
Ok(WorkerState::Done)
}
}
}
}
async fn wait_for_work(&mut self, _must_exit: &watch::Receiver<bool>) -> WorkerState {
unreachable!()
}
}
// ---- ---- ----
// SECOND KIND OF REPAIR: SCRUBBING THE DATASTORE
// This is significantly more complex than the process above,
// as it is a continuously-running task that triggers automatically
// every SCRUB_INTERVAL, but can also be triggered manually
// and whose parameter (esp. speed) can be controlled at runtime.
// ---- ---- ----
pub struct ScrubWorker {
manager: Arc<BlockManager>,
rx_cmd: mpsc::Receiver<ScrubWorkerCommand>,
work: ScrubWorkerState,
tranquilizer: Tranquilizer,
persister: Persister<ScrubWorkerPersisted>,
persisted: ScrubWorkerPersisted,
}
#[derive(Serialize, Deserialize)]
struct ScrubWorkerPersisted {
tranquility: u32,
time_last_complete_scrub: u64,
corruptions_detected: u64,
}
enum ScrubWorkerState {
Running(BlockStoreIterator),
Paused(BlockStoreIterator, u64), // u64 = time when to resume scrub
Finished,
}
impl Default for ScrubWorkerState {
fn default() -> Self {
ScrubWorkerState::Finished
}
}
#[derive(Debug)]
pub enum ScrubWorkerCommand {
Start,
Pause(Duration),
Resume,
Cancel,
SetTranquility(u32),
}
impl ScrubWorker {
pub fn new(manager: Arc<BlockManager>, rx_cmd: mpsc::Receiver<ScrubWorkerCommand>) -> Self {
let persister = Persister::new(&manager.system.metadata_dir, "scrub_info");
let persisted = match persister.load() {
Ok(v) => v,
Err(_) => ScrubWorkerPersisted {
time_last_complete_scrub: 0,
tranquility: INITIAL_SCRUB_TRANQUILITY,
corruptions_detected: 0,
},
};
Self {
manager,
rx_cmd,
work: ScrubWorkerState::Finished,
tranquilizer: Tranquilizer::new(30),
persister,
persisted,
}
}
async fn handle_cmd(&mut self, cmd: ScrubWorkerCommand) {
match cmd {
ScrubWorkerCommand::Start => {
self.work = match std::mem::take(&mut self.work) {
ScrubWorkerState::Finished => {
let iterator = BlockStoreIterator::new(&self.manager);
ScrubWorkerState::Running(iterator)
}
work => {
error!("Cannot start scrub worker: already running!");
work
}
};
}
ScrubWorkerCommand::Pause(dur) => {
self.work = match std::mem::take(&mut self.work) {
ScrubWorkerState::Running(it) | ScrubWorkerState::Paused(it, _) => {
ScrubWorkerState::Paused(it, now_msec() + dur.as_millis() as u64)
}
work => {
error!("Cannot pause scrub worker: not running!");
work
}
};
}
ScrubWorkerCommand::Resume => {
self.work = match std::mem::take(&mut self.work) {
ScrubWorkerState::Paused(it, _) => ScrubWorkerState::Running(it),
work => {
error!("Cannot resume scrub worker: not paused!");
work
}
};
}
ScrubWorkerCommand::Cancel => {
self.work = match std::mem::take(&mut self.work) {
ScrubWorkerState::Running(_) | ScrubWorkerState::Paused(_, _) => {
ScrubWorkerState::Finished
}
work => {
error!("Cannot cancel scrub worker: not running!");
work
}
}
}
ScrubWorkerCommand::SetTranquility(t) => {
self.persisted.tranquility = t;
if let Err(e) = self.persister.save_async(&self.persisted).await {
error!("Could not save new tranquilitiy value: {}", e);
}
}
}
}
}
#[async_trait]
impl Worker for ScrubWorker {
fn name(&self) -> String {
"Block scrub worker".into()
}
fn info(&self) -> Option<String> {
let s = match &self.work {
ScrubWorkerState::Running(bsi) => format!(
"{:.2}% done (tranquility = {})",
bsi.progress() * 100.,
self.persisted.tranquility
),
ScrubWorkerState::Paused(bsi, rt) => {
format!(
"Paused, {:.2}% done, resumes at {}",
bsi.progress() * 100.,
msec_to_rfc3339(*rt)
)
}
ScrubWorkerState::Finished => format!(
"Last completed scrub: {}",
msec_to_rfc3339(self.persisted.time_last_complete_scrub)
),
};
Some(format!(
"{} ; corruptions detected: {}",
s, self.persisted.corruptions_detected
))
}
async fn work(&mut self, _must_exit: &mut watch::Receiver<bool>) -> Result<WorkerState, Error> {
match self.rx_cmd.try_recv() {
Ok(cmd) => self.handle_cmd(cmd).await,
Err(mpsc::error::TryRecvError::Disconnected) => return Ok(WorkerState::Done),
Err(mpsc::error::TryRecvError::Empty) => (),
};
match &mut self.work {
ScrubWorkerState::Running(bsi) => {
self.tranquilizer.reset();
if let Some(hash) = bsi.next().await? {
match self.manager.read_block(&hash).await {
Err(Error::CorruptData(_)) => {
error!("Found corrupt data block during scrub: {:?}", hash);
self.persisted.corruptions_detected += 1;
self.persister.save_async(&self.persisted).await?;
}
Err(e) => return Err(e),
_ => (),
};
Ok(self
.tranquilizer
.tranquilize_worker(self.persisted.tranquility))
} else {
self.persisted.time_last_complete_scrub = now_msec();
self.persister.save_async(&self.persisted).await?;
self.work = ScrubWorkerState::Finished;
self.tranquilizer.clear();
Ok(WorkerState::Idle)
}
}
_ => Ok(WorkerState::Idle),
}
}
async fn wait_for_work(&mut self, _must_exit: &watch::Receiver<bool>) -> WorkerState {
let (wait_until, command) = match &self.work {
ScrubWorkerState::Running(_) => return WorkerState::Busy,
ScrubWorkerState::Paused(_, resume_time) => (*resume_time, ScrubWorkerCommand::Resume),
ScrubWorkerState::Finished => (
self.persisted.time_last_complete_scrub + SCRUB_INTERVAL.as_millis() as u64,
ScrubWorkerCommand::Start,
),
};
let now = now_msec();
if now >= wait_until {
self.handle_cmd(command).await;
return WorkerState::Busy;
}
let delay = Duration::from_millis(wait_until - now);
select! {
_ = tokio::time::sleep(delay) => self.handle_cmd(command).await,
cmd = self.rx_cmd.recv() => if let Some(cmd) = cmd {
self.handle_cmd(cmd).await;
} else {
return WorkerState::Done;
}
}
match &self.work {
ScrubWorkerState::Running(_) => WorkerState::Busy,
_ => WorkerState::Idle,
}
}
}
// ---- ---- ----
// UTILITY FOR ENUMERATING THE BLOCK STORE
// ---- ---- ----
struct BlockStoreIterator {
path: Vec<ReadingDir>,
}
enum ReadingDir {
Pending(PathBuf),
Read {
subpaths: Vec<fs::DirEntry>,
pos: usize,
},
}
impl BlockStoreIterator {
fn new(manager: &BlockManager) -> Self {
let root_dir = manager.data_dir.clone();
Self {
path: vec![ReadingDir::Pending(root_dir)],
}
}
/// Returns progress done, between 0 and 1
fn progress(&self) -> f32 {
if self.path.is_empty() {
1.0
} else {
let mut ret = 0.0;
let mut next_div = 1;
for p in self.path.iter() {
match p {
ReadingDir::Pending(_) => break,
ReadingDir::Read { subpaths, pos } => {
next_div *= subpaths.len();
ret += ((*pos - 1) as f32) / (next_div as f32);
}
}
}
ret
}
}
async fn next(&mut self) -> Result<Option<Hash>, Error> {
loop {
let last_path = match self.path.last_mut() {
None => return Ok(None),
Some(lp) => lp,
};
if let ReadingDir::Pending(path) = last_path {
let mut reader = fs::read_dir(&path).await?;
let mut subpaths = vec![];
while let Some(ent) = reader.next_entry().await? {
subpaths.push(ent);
}
*last_path = ReadingDir::Read { subpaths, pos: 0 };
}
let (subpaths, pos) = match *last_path {
ReadingDir::Read {
ref subpaths,
ref mut pos,
} => (subpaths, pos),
ReadingDir::Pending(_) => unreachable!(),
};
let data_dir_ent = match subpaths.get(*pos) {
None => {
self.path.pop();
continue;
}
Some(ent) => {
*pos += 1;
ent
}
};
let name = data_dir_ent.file_name();
let name = if let Ok(n) = name.into_string() {
n
} else {
continue;
};
let ent_type = data_dir_ent.file_type().await?;
let name = name.strip_suffix(".zst").unwrap_or(&name);
if name.len() == 2 && hex::decode(&name).is_ok() && ent_type.is_dir() {
let path = data_dir_ent.path();
self.path.push(ReadingDir::Pending(path));
} else if name.len() == 64 {
if let Ok(h) = hex::decode(&name) {
let mut hash = [0u8; 32];
hash.copy_from_slice(&h);
return Ok(Some(hash.into()));
}
}
}
}
}
+589
View File
@@ -0,0 +1,589 @@
use std::collections::HashSet;
use std::convert::TryInto;
use std::sync::{Arc, Mutex};
use std::time::Duration;
use arc_swap::ArcSwap;
use async_trait::async_trait;
use serde::{Deserialize, Serialize};
use tokio::select;
use tokio::sync::{watch, Notify};
use opentelemetry::{
trace::{FutureExt as OtelFutureExt, TraceContextExt, Tracer},
Context, KeyValue,
};
use garage_db as db;
use garage_db::counted_tree_hack::CountedTree;
use garage_util::background::*;
use garage_util::data::*;
use garage_util::error::*;
use garage_util::metrics::RecordDuration;
use garage_util::persister::Persister;
use garage_util::time::*;
use garage_util::tranquilizer::Tranquilizer;
use garage_rpc::system::System;
use garage_rpc::*;
use garage_table::replication::TableReplication;
use crate::manager::*;
// The delay between the time where a resync operation fails
// and the time when it is retried, with exponential backoff
// (multiplied by 2, 4, 8, 16, etc. for every consecutive failure).
pub(crate) const RESYNC_RETRY_DELAY: Duration = Duration::from_secs(60);
// The minimum retry delay is 60 seconds = 1 minute
// The maximum retry delay is 60 seconds * 2^6 = 60 seconds << 6 = 64 minutes (~1 hour)
pub(crate) const RESYNC_RETRY_DELAY_MAX_BACKOFF_POWER: u64 = 6;
// No more than 4 resync workers can be running in the system
pub(crate) const MAX_RESYNC_WORKERS: usize = 4;
// Resync tranquility is initially set to 2, but can be changed in the CLI
// and the updated version is persisted over Garage restarts
const INITIAL_RESYNC_TRANQUILITY: u32 = 2;
pub struct BlockResyncManager {
pub(crate) queue: CountedTree,
pub(crate) notify: Notify,
pub(crate) errors: CountedTree,
busy_set: BusySet,
persister: Persister<ResyncPersistedConfig>,
persisted: ArcSwap<ResyncPersistedConfig>,
}
#[derive(Serialize, Deserialize, Clone, Copy)]
struct ResyncPersistedConfig {
n_workers: usize,
tranquility: u32,
}
enum ResyncIterResult {
BusyDidSomething,
BusyDidNothing,
IdleFor(Duration),
}
type BusySet = Arc<Mutex<HashSet<Vec<u8>>>>;
struct BusyBlock {
time_bytes: Vec<u8>,
hash_bytes: Vec<u8>,
busy_set: BusySet,
}
impl BlockResyncManager {
pub(crate) fn new(db: &db::Db, system: &System) -> Self {
let queue = db
.open_tree("block_local_resync_queue")
.expect("Unable to open block_local_resync_queue tree");
let queue = CountedTree::new(queue).expect("Could not count block_local_resync_queue");
let errors = db
.open_tree("block_local_resync_errors")
.expect("Unable to open block_local_resync_errors tree");
let errors = CountedTree::new(errors).expect("Could not count block_local_resync_errors");
let persister = Persister::new(&system.metadata_dir, "resync_cfg");
let persisted = match persister.load() {
Ok(v) => v,
Err(_) => ResyncPersistedConfig {
n_workers: 1,
tranquility: INITIAL_RESYNC_TRANQUILITY,
},
};
Self {
queue,
notify: Notify::new(),
errors,
busy_set: Arc::new(Mutex::new(HashSet::new())),
persister,
persisted: ArcSwap::new(Arc::new(persisted)),
}
}
/// Get lenght of resync queue
pub fn queue_len(&self) -> Result<usize, Error> {
// This currently can't return an error because the CountedTree hack
// doesn't error on .len(), but this will change when we remove the hack
// (hopefully someday!)
Ok(self.queue.len())
}
/// Get number of blocks that have an error
pub fn errors_len(&self) -> Result<usize, Error> {
// (see queue_len comment)
Ok(self.errors.len())
}
// ---- Resync loop ----
// This part manages a queue of blocks that need to be
// "resynchronized", i.e. that need to have a check that
// they are at present if we need them, or that they are
// deleted once the garbage collection delay has passed.
//
// Here are some explanations on how the resync queue works.
// There are two Sled trees that are used to have information
// about the status of blocks that need to be resynchronized:
//
// - resync.queue: a tree that is ordered first by a timestamp
// (in milliseconds since Unix epoch) that is the time at which
// the resync must be done, and second by block hash.
// The key in this tree is just:
// concat(timestamp (8 bytes), hash (32 bytes))
// The value is the same 32-byte hash.
//
// - resync.errors: a tree that indicates for each block
// if the last resync resulted in an error, and if so,
// the following two informations (see the ErrorCounter struct):
// - how many consecutive resync errors for this block?
// - when was the last try?
// These two informations are used to implement an
// exponential backoff retry strategy.
// The key in this tree is the 32-byte hash of the block,
// and the value is the encoded ErrorCounter value.
//
// We need to have these two trees, because the resync queue
// is not just a queue of items to process, but a set of items
// that are waiting a specific delay until we can process them
// (the delay being necessary both internally for the exponential
// backoff strategy, and exposed as a parameter when adding items
// to the queue, e.g. to wait until the GC delay has passed).
// This is why we need one tree ordered by time, and one
// ordered by identifier of item to be processed (block hash).
//
// When the worker wants to process an item it takes from
// resync.queue, it checks in resync.errors that if there is an
// exponential back-off delay to await, it has passed before we
// process the item. If not, the item in the queue is skipped
// (but added back for later processing after the time of the
// delay).
//
// An alternative that would have seemed natural is to
// only add items to resync.queue with a processing time that is
// after the delay, but there are several issues with this:
// - This requires to synchronize updates to resync.queue and
// resync.errors (with the current model, there is only one thread,
// the worker thread, that accesses resync.errors,
// so no need to synchronize) by putting them both in a lock.
// This would mean that block_incref might need to take a lock
// before doing its thing, meaning it has much more chances of
// not completing successfully if something bad happens to Garage.
// Currently Garage is not able to recover from block_incref that
// doesn't complete successfully, because it is necessary to ensure
// the consistency between the state of the block manager and
// information in the BlockRef table.
// - If a resync fails, we put that block in the resync.errors table,
// and also add it back to resync.queue to be processed after
// the exponential back-off delay,
// but maybe the block is already scheduled to be resynced again
// at another time that is before the exponential back-off delay,
// and we have no way to check that easily. This means that
// in all cases, we need to check the resync.errors table
// in the resync loop at the time when a block is popped from
// the resync.queue.
// Overall, the current design is therefore simpler and more robust
// because it tolerates inconsistencies between the resync.queue
// and resync.errors table (items being scheduled in resync.queue
// for times that are earlier than the exponential back-off delay
// is a natural condition that is handled properly).
pub(crate) fn put_to_resync(&self, hash: &Hash, delay: Duration) -> db::Result<()> {
let when = now_msec() + delay.as_millis() as u64;
self.put_to_resync_at(hash, when)
}
pub(crate) fn put_to_resync_at(&self, hash: &Hash, when: u64) -> db::Result<()> {
trace!("Put resync_queue: {} {:?}", when, hash);
let mut key = u64::to_be_bytes(when).to_vec();
key.extend(hash.as_ref());
self.queue.insert(key, hash.as_ref())?;
self.notify.notify_waiters();
Ok(())
}
async fn resync_iter(&self, manager: &BlockManager) -> Result<ResyncIterResult, db::Error> {
if let Some(block) = self.get_block_to_resync()? {
let time_msec = u64::from_be_bytes(block.time_bytes[0..8].try_into().unwrap());
let now = now_msec();
if now >= time_msec {
let hash = Hash::try_from(&block.hash_bytes[..]).unwrap();
if let Some(ec) = self.errors.get(hash.as_slice())? {
let ec = ErrorCounter::decode(&ec);
if now < ec.next_try() {
// if next retry after an error is not yet,
// don't do resync and return early, but still
// make sure the item is still in queue at expected time
self.put_to_resync_at(&hash, ec.next_try())?;
// ec.next_try() > now >= time_msec, so this remove
// is not removing the one we added just above
// (we want to do the remove after the insert to ensure
// that the item is not lost if we crash in-between)
self.queue.remove(&block.time_bytes)?;
return Ok(ResyncIterResult::BusyDidNothing);
}
}
let tracer = opentelemetry::global::tracer("garage");
let trace_id = gen_uuid();
let span = tracer
.span_builder("Resync block")
.with_trace_id(
opentelemetry::trace::TraceId::from_hex(&hex::encode(
&trace_id.as_slice()[..16],
))
.unwrap(),
)
.with_attributes(vec![KeyValue::new("block", format!("{:?}", hash))])
.start(&tracer);
let res = self
.resync_block(manager, &hash)
.with_context(Context::current_with_span(span))
.bound_record_duration(&manager.metrics.resync_duration)
.await;
manager.metrics.resync_counter.add(1);
if let Err(e) = &res {
manager.metrics.resync_error_counter.add(1);
warn!("Error when resyncing {:?}: {}", hash, e);
let err_counter = match self.errors.get(hash.as_slice())? {
Some(ec) => ErrorCounter::decode(&ec).add1(now + 1),
None => ErrorCounter::new(now + 1),
};
self.errors.insert(hash.as_slice(), err_counter.encode())?;
self.put_to_resync_at(&hash, err_counter.next_try())?;
// err_counter.next_try() >= now + 1 > now,
// the entry we remove from the queue is not
// the entry we inserted with put_to_resync_at
self.queue.remove(&block.time_bytes)?;
} else {
self.errors.remove(hash.as_slice())?;
self.queue.remove(&block.time_bytes)?;
}
Ok(ResyncIterResult::BusyDidSomething)
} else {
Ok(ResyncIterResult::IdleFor(Duration::from_millis(
time_msec - now,
)))
}
} else {
// Here we wait either for a notification that an item has been
// added to the queue, or for a constant delay of 10 secs to expire.
// The delay avoids a race condition where the notification happens
// between the time we checked the queue and the first poll
// to resync_notify.notified(): if that happens, we'll just loop
// back 10 seconds later, which is fine.
Ok(ResyncIterResult::IdleFor(Duration::from_secs(10)))
}
}
fn get_block_to_resync(&self) -> Result<Option<BusyBlock>, db::Error> {
let mut busy = self.busy_set.lock().unwrap();
for it in self.queue.iter()? {
let (time_bytes, hash_bytes) = it?;
if !busy.contains(&time_bytes) {
busy.insert(time_bytes.clone());
return Ok(Some(BusyBlock {
time_bytes,
hash_bytes,
busy_set: self.busy_set.clone(),
}));
}
}
Ok(None)
}
async fn resync_block(&self, manager: &BlockManager, hash: &Hash) -> Result<(), Error> {
let BlockStatus { exists, needed } = manager.check_block_status(hash).await?;
if exists != needed.is_needed() || exists != needed.is_nonzero() {
debug!(
"Resync block {:?}: exists {}, nonzero rc {}, deletable {}",
hash,
exists,
needed.is_nonzero(),
needed.is_deletable(),
);
}
if exists && needed.is_deletable() {
info!("Resync block {:?}: offloading and deleting", hash);
let mut who = manager.replication.write_nodes(hash);
if who.len() < manager.replication.write_quorum() {
return Err(Error::Message("Not trying to offload block because we don't have a quorum of nodes to write to".to_string()));
}
who.retain(|id| *id != manager.system.id);
let who_needs_resps = manager
.system
.rpc
.call_many(
&manager.endpoint,
&who,
BlockRpc::NeedBlockQuery(*hash),
RequestStrategy::with_priority(PRIO_BACKGROUND),
)
.await?;
let mut need_nodes = vec![];
for (node, needed) in who_needs_resps {
match needed.err_context("NeedBlockQuery RPC")? {
BlockRpc::NeedBlockReply(needed) => {
if needed {
need_nodes.push(node);
}
}
m => {
return Err(Error::unexpected_rpc_message(m));
}
}
}
if !need_nodes.is_empty() {
trace!(
"Block {:?} needed by {} nodes, sending",
hash,
need_nodes.len()
);
for node in need_nodes.iter() {
manager
.metrics
.resync_send_counter
.add(1, &[KeyValue::new("to", format!("{:?}", node))]);
}
let block = manager.read_block(hash).await?;
let (header, bytes) = block.into_parts();
let put_block_message = Req::new(BlockRpc::PutBlock {
hash: *hash,
header,
})?
.with_stream_from_buffer(bytes);
manager
.system
.rpc
.try_call_many(
&manager.endpoint,
&need_nodes[..],
put_block_message,
RequestStrategy::with_priority(PRIO_BACKGROUND)
.with_quorum(need_nodes.len()),
)
.await
.err_context("PutBlock RPC")?;
}
info!(
"Deleting unneeded block {:?}, offload finished ({} / {})",
hash,
need_nodes.len(),
who.len()
);
manager.delete_if_unneeded(hash).await?;
manager.rc.clear_deleted_block_rc(hash)?;
}
if needed.is_nonzero() && !exists {
info!(
"Resync block {:?}: fetching absent but needed block (refcount > 0)",
hash
);
let block_data = manager.rpc_get_raw_block(hash, None).await?;
manager.metrics.resync_recv_counter.add(1);
manager.write_block(hash, &block_data).await?;
}
Ok(())
}
async fn update_persisted(
&self,
update: impl Fn(&mut ResyncPersistedConfig),
) -> Result<(), Error> {
let mut cfg: ResyncPersistedConfig = *self.persisted.load().as_ref();
update(&mut cfg);
self.persister.save_async(&cfg).await?;
self.persisted.store(Arc::new(cfg));
self.notify.notify_waiters();
Ok(())
}
pub async fn set_n_workers(&self, n_workers: usize) -> Result<(), Error> {
if !(1..=MAX_RESYNC_WORKERS).contains(&n_workers) {
return Err(Error::Message(format!(
"Invalid number of resync workers, must be between 1 and {}",
MAX_RESYNC_WORKERS
)));
}
self.update_persisted(|cfg| cfg.n_workers = n_workers).await
}
pub async fn set_tranquility(&self, tranquility: u32) -> Result<(), Error> {
self.update_persisted(|cfg| cfg.tranquility = tranquility)
.await
}
}
impl Drop for BusyBlock {
fn drop(&mut self) {
let mut busy = self.busy_set.lock().unwrap();
busy.remove(&self.time_bytes);
}
}
pub(crate) struct ResyncWorker {
index: usize,
manager: Arc<BlockManager>,
tranquilizer: Tranquilizer,
next_delay: Duration,
}
impl ResyncWorker {
pub(crate) fn new(index: usize, manager: Arc<BlockManager>) -> Self {
Self {
index,
manager,
tranquilizer: Tranquilizer::new(30),
next_delay: Duration::from_secs(10),
}
}
}
#[async_trait]
impl Worker for ResyncWorker {
fn name(&self) -> String {
format!("Block resync worker #{}", self.index + 1)
}
fn info(&self) -> Option<String> {
let persisted = self.manager.resync.persisted.load();
if self.index >= persisted.n_workers {
return Some("(unused)".into());
}
let mut ret = vec![];
ret.push(format!("tranquility = {}", persisted.tranquility));
let qlen = self.manager.resync.queue_len().unwrap_or(0);
if qlen > 0 {
ret.push(format!("{} blocks in queue", qlen));
}
let elen = self.manager.resync.errors_len().unwrap_or(0);
if elen > 0 {
ret.push(format!("{} blocks in error state", elen));
}
Some(ret.join(", "))
}
async fn work(&mut self, _must_exit: &mut watch::Receiver<bool>) -> Result<WorkerState, Error> {
if self.index >= self.manager.resync.persisted.load().n_workers {
return Ok(WorkerState::Idle);
}
self.tranquilizer.reset();
match self.manager.resync.resync_iter(&self.manager).await {
Ok(ResyncIterResult::BusyDidSomething) => Ok(self
.tranquilizer
.tranquilize_worker(self.manager.resync.persisted.load().tranquility)),
Ok(ResyncIterResult::BusyDidNothing) => Ok(WorkerState::Busy),
Ok(ResyncIterResult::IdleFor(delay)) => {
self.next_delay = delay;
Ok(WorkerState::Idle)
}
Err(e) => {
// The errors that we have here are only Sled errors
// We don't really know how to handle them so just ¯\_(ツ)_/¯
// (there is kind of an assumption that Sled won't error on us,
// if it does there is not much we can do -- TODO should we just panic?)
// Here we just give the error to the worker manager,
// it will print it to the logs and increment a counter
Err(e.into())
}
}
}
async fn wait_for_work(&mut self, _must_exit: &watch::Receiver<bool>) -> WorkerState {
while self.index >= self.manager.resync.persisted.load().n_workers {
self.manager.resync.notify.notified().await
}
select! {
_ = tokio::time::sleep(self.next_delay) => (),
_ = self.manager.resync.notify.notified() => (),
};
WorkerState::Busy
}
}
/// Counts the number of errors when resyncing a block,
/// and the time of the last try.
/// Used to implement exponential backoff.
#[derive(Clone, Copy, Debug)]
struct ErrorCounter {
errors: u64,
last_try: u64,
}
impl ErrorCounter {
fn new(now: u64) -> Self {
Self {
errors: 1,
last_try: now,
}
}
fn decode(data: &[u8]) -> Self {
Self {
errors: u64::from_be_bytes(data[0..8].try_into().unwrap()),
last_try: u64::from_be_bytes(data[8..16].try_into().unwrap()),
}
}
fn encode(&self) -> Vec<u8> {
[
u64::to_be_bytes(self.errors),
u64::to_be_bytes(self.last_try),
]
.concat()
}
fn add1(self, now: u64) -> Self {
Self {
errors: self.errors + 1,
last_try: now,
}
}
fn delay_msec(&self) -> u64 {
(RESYNC_RETRY_DELAY.as_millis() as u64)
<< std::cmp::min(self.errors - 1, RESYNC_RETRY_DELAY_MAX_BACKOFF_POWER)
}
fn next_try(&self) -> u64 {
self.last_try + self.delay_msec()
}
}
+39
View File
@@ -0,0 +1,39 @@
[package]
name = "garage_db"
version = "0.8.0"
authors = ["Alex Auvolat <alex@adnab.me>"]
edition = "2018"
license = "AGPL-3.0"
description = "Abstraction over multiple key/value storage engines that supports transactions"
repository = "https://git.deuxfleurs.fr/Deuxfleurs/garage"
readme = "../../README.md"
[lib]
path = "lib.rs"
[[bin]]
name = "convert"
path = "bin/convert.rs"
required-features = ["cli"]
[dependencies]
err-derive = "0.3"
hexdump = "0.1"
tracing = "0.1.30"
heed = { version = "0.11", default-features = false, features = ["lmdb"], optional = true }
rusqlite = { version = "0.27", optional = true }
sled = { version = "0.34", optional = true }
# cli deps
clap = { version = "3.1.18", optional = true, features = ["derive", "env"] }
pretty_env_logger = { version = "0.4", optional = true }
[dev-dependencies]
mktemp = "0.4"
[features]
bundled-libs = [ "rusqlite/bundled" ]
cli = ["clap", "pretty_env_logger"]
lmdb = [ "heed" ]
sqlite = [ "rusqlite" ]
+69
View File
@@ -0,0 +1,69 @@
use std::path::PathBuf;
use garage_db::*;
use clap::Parser;
/// K2V command line interface
#[derive(Parser, Debug)]
#[clap(author, version, about, long_about = None)]
struct Args {
/// Input DB path
#[clap(short = 'i')]
input_path: PathBuf,
/// Input DB engine
#[clap(short = 'a')]
input_engine: String,
/// Output DB path
#[clap(short = 'o')]
output_path: PathBuf,
/// Output DB engine
#[clap(short = 'b')]
output_engine: String,
}
fn main() {
let args = Args::parse();
pretty_env_logger::init();
match do_conversion(args) {
Ok(()) => println!("Success!"),
Err(e) => eprintln!("Error: {}", e),
}
}
fn do_conversion(args: Args) -> Result<()> {
let input = open_db(args.input_path, args.input_engine)?;
let output = open_db(args.output_path, args.output_engine)?;
output.import(&input)?;
Ok(())
}
fn open_db(path: PathBuf, engine: String) -> Result<Db> {
match engine.as_str() {
"sled" => {
let db = sled_adapter::sled::Config::default().path(&path).open()?;
Ok(sled_adapter::SledDb::init(db))
}
"sqlite" | "sqlite3" | "rusqlite" => {
let db = sqlite_adapter::rusqlite::Connection::open(&path)?;
Ok(sqlite_adapter::SqliteDb::init(db))
}
"lmdb" | "heed" => {
std::fs::create_dir_all(&path).map_err(|e| {
Error(format!("Unable to create LMDB data directory: {}", e).into())
})?;
let map_size = lmdb_adapter::recommended_map_size();
let db = lmdb_adapter::heed::EnvOpenOptions::new()
.max_dbs(100)
.map_size(map_size)
.open(&path)
.unwrap();
Ok(lmdb_adapter::LmdbDb::init(db))
}
e => Err(Error(format!("Invalid DB engine: {}", e).into())),
}
}
+127
View File
@@ -0,0 +1,127 @@
//! This hack allows a db tree to keep in RAM a counter of the number of entries
//! it contains, which is used to call .len() on it. This is usefull only for
//! the sled backend where .len() otherwise would have to traverse the whole
//! tree to count items. For sqlite and lmdb, this is mostly useless (but
//! hopefully not harmfull!). Note that a CountedTree cannot be part of a
//! transaction.
use std::sync::{
atomic::{AtomicUsize, Ordering},
Arc,
};
use crate::{Result, Tree, TxError, Value, ValueIter};
#[derive(Clone)]
pub struct CountedTree(Arc<CountedTreeInternal>);
struct CountedTreeInternal {
tree: Tree,
len: AtomicUsize,
}
impl CountedTree {
pub fn new(tree: Tree) -> Result<Self> {
let len = tree.len()?;
Ok(Self(Arc::new(CountedTreeInternal {
tree,
len: AtomicUsize::new(len),
})))
}
pub fn len(&self) -> usize {
self.0.len.load(Ordering::SeqCst)
}
pub fn is_empty(&self) -> bool {
self.len() == 0
}
pub fn get<K: AsRef<[u8]>>(&self, key: K) -> Result<Option<Value>> {
self.0.tree.get(key)
}
pub fn first(&self) -> Result<Option<(Value, Value)>> {
self.0.tree.first()
}
pub fn iter(&self) -> Result<ValueIter<'_>> {
self.0.tree.iter()
}
// ---- writing functions ----
pub fn insert<K, V>(&self, key: K, value: V) -> Result<Option<Value>>
where
K: AsRef<[u8]>,
V: AsRef<[u8]>,
{
let old_val = self.0.tree.insert(key, value)?;
if old_val.is_none() {
self.0.len.fetch_add(1, Ordering::SeqCst);
}
Ok(old_val)
}
pub fn remove<K: AsRef<[u8]>>(&self, key: K) -> Result<Option<Value>> {
let old_val = self.0.tree.remove(key)?;
if old_val.is_some() {
self.0.len.fetch_sub(1, Ordering::SeqCst);
}
Ok(old_val)
}
pub fn compare_and_swap<K, OV, NV>(
&self,
key: K,
expected_old: Option<OV>,
new: Option<NV>,
) -> Result<bool>
where
K: AsRef<[u8]>,
OV: AsRef<[u8]>,
NV: AsRef<[u8]>,
{
let old_some = expected_old.is_some();
let new_some = new.is_some();
let tx_res = self.0.tree.db().transaction(|mut tx| {
let old_val = tx.get(&self.0.tree, &key)?;
let is_same = match (&old_val, &expected_old) {
(None, None) => true,
(Some(x), Some(y)) if x == y.as_ref() => true,
_ => false,
};
if is_same {
match &new {
Some(v) => {
tx.insert(&self.0.tree, &key, v)?;
}
None => {
tx.remove(&self.0.tree, &key)?;
}
}
tx.commit(())
} else {
tx.abort(())
}
});
match tx_res {
Ok(()) => {
match (old_some, new_some) {
(false, true) => {
self.0.len.fetch_add(1, Ordering::SeqCst);
}
(true, false) => {
self.0.len.fetch_sub(1, Ordering::SeqCst);
}
_ => (),
}
Ok(true)
}
Err(TxError::Abort(())) => Ok(false),
Err(TxError::Db(e)) => Err(e),
}
}
}
+416
View File
@@ -0,0 +1,416 @@
#[macro_use]
#[cfg(feature = "sqlite")]
extern crate tracing;
#[cfg(not(any(feature = "lmdb", feature = "sled", feature = "sqlite")))]
compile_error!("Must activate the Cargo feature for at least one DB engine: lmdb, sled or sqlite.");
#[cfg(feature = "lmdb")]
pub mod lmdb_adapter;
#[cfg(feature = "sled")]
pub mod sled_adapter;
#[cfg(feature = "sqlite")]
pub mod sqlite_adapter;
pub mod counted_tree_hack;
#[cfg(test)]
pub mod test;
use core::ops::{Bound, RangeBounds};
use std::borrow::Cow;
use std::cell::Cell;
use std::sync::Arc;
use err_derive::Error;
#[derive(Clone)]
pub struct Db(pub(crate) Arc<dyn IDb>);
pub struct Transaction<'a>(&'a mut dyn ITx);
#[derive(Clone)]
pub struct Tree(Arc<dyn IDb>, usize);
pub type Value = Vec<u8>;
pub type ValueIter<'a> = Box<dyn std::iter::Iterator<Item = Result<(Value, Value)>> + 'a>;
pub type TxValueIter<'a> = Box<dyn std::iter::Iterator<Item = TxOpResult<(Value, Value)>> + 'a>;
// ----
#[derive(Debug, Error)]
#[error(display = "{}", _0)]
pub struct Error(pub Cow<'static, str>);
pub type Result<T> = std::result::Result<T, Error>;
#[derive(Debug, Error)]
#[error(display = "{}", _0)]
pub struct TxOpError(pub(crate) Error);
pub type TxOpResult<T> = std::result::Result<T, TxOpError>;
pub enum TxError<E> {
Abort(E),
Db(Error),
}
pub type TxResult<R, E> = std::result::Result<R, TxError<E>>;
impl<E> From<TxOpError> for TxError<E> {
fn from(e: TxOpError) -> TxError<E> {
TxError::Db(e.0)
}
}
pub fn unabort<R, E>(res: TxResult<R, E>) -> TxOpResult<std::result::Result<R, E>> {
match res {
Ok(v) => Ok(Ok(v)),
Err(TxError::Abort(e)) => Ok(Err(e)),
Err(TxError::Db(e)) => Err(TxOpError(e)),
}
}
// ----
impl Db {
pub fn engine(&self) -> String {
self.0.engine()
}
pub fn open_tree<S: AsRef<str>>(&self, name: S) -> Result<Tree> {
let tree_id = self.0.open_tree(name.as_ref())?;
Ok(Tree(self.0.clone(), tree_id))
}
pub fn list_trees(&self) -> Result<Vec<String>> {
self.0.list_trees()
}
pub fn transaction<R, E, F>(&self, fun: F) -> TxResult<R, E>
where
F: Fn(Transaction<'_>) -> TxResult<R, E>,
{
let f = TxFn {
function: fun,
result: Cell::new(None),
};
let tx_res = self.0.transaction(&f);
let ret = f
.result
.into_inner()
.expect("Transaction did not store result");
match tx_res {
Ok(()) => {
assert!(matches!(ret, Ok(_)));
ret
}
Err(TxError::Abort(())) => {
assert!(matches!(ret, Err(TxError::Abort(_))));
ret
}
Err(TxError::Db(e2)) => match ret {
// Ok was stored -> the error occured when finalizing
// transaction
Ok(_) => Err(TxError::Db(e2)),
// An error was already stored: that's the one we want to
// return
Err(TxError::Db(e)) => Err(TxError::Db(e)),
_ => unreachable!(),
},
}
}
pub fn import(&self, other: &Db) -> Result<()> {
let existing_trees = self.list_trees()?;
if !existing_trees.is_empty() {
return Err(Error(
format!(
"destination database already contains data: {:?}",
existing_trees
)
.into(),
));
}
let tree_names = other.list_trees()?;
for name in tree_names {
let tree = self.open_tree(&name)?;
if tree.len()? > 0 {
return Err(Error(format!("tree {} already contains data", name).into()));
}
let ex_tree = other.open_tree(&name)?;
let tx_res = self.transaction(|mut tx| {
let mut i = 0;
for item in ex_tree.iter().map_err(TxError::Abort)? {
let (k, v) = item.map_err(TxError::Abort)?;
tx.insert(&tree, k, v)?;
i += 1;
if i % 1000 == 0 {
println!("{}: imported {}", name, i);
}
}
tx.commit(i)
});
let total = match tx_res {
Err(TxError::Db(e)) => return Err(e),
Err(TxError::Abort(e)) => return Err(e),
Ok(x) => x,
};
println!("{}: finished importing, {} items", name, total);
}
Ok(())
}
}
#[allow(clippy::len_without_is_empty)]
impl Tree {
#[inline]
pub fn db(&self) -> Db {
Db(self.0.clone())
}
#[inline]
pub fn get<T: AsRef<[u8]>>(&self, key: T) -> Result<Option<Value>> {
self.0.get(self.1, key.as_ref())
}
#[inline]
pub fn len(&self) -> Result<usize> {
self.0.len(self.1)
}
#[inline]
pub fn first(&self) -> Result<Option<(Value, Value)>> {
self.iter()?.next().transpose()
}
#[inline]
pub fn get_gt<T: AsRef<[u8]>>(&self, from: T) -> Result<Option<(Value, Value)>> {
self.range((Bound::Excluded(from), Bound::Unbounded))?
.next()
.transpose()
}
/// Returns the old value if there was one
#[inline]
pub fn insert<T: AsRef<[u8]>, U: AsRef<[u8]>>(
&self,
key: T,
value: U,
) -> Result<Option<Value>> {
self.0.insert(self.1, key.as_ref(), value.as_ref())
}
/// Returns the old value if there was one
#[inline]
pub fn remove<T: AsRef<[u8]>>(&self, key: T) -> Result<Option<Value>> {
self.0.remove(self.1, key.as_ref())
}
/// Clears all values from the tree
#[inline]
pub fn clear(&self) -> Result<()> {
self.0.clear(self.1)
}
#[inline]
pub fn iter(&self) -> Result<ValueIter<'_>> {
self.0.iter(self.1)
}
#[inline]
pub fn iter_rev(&self) -> Result<ValueIter<'_>> {
self.0.iter_rev(self.1)
}
#[inline]
pub fn range<K, R>(&self, range: R) -> Result<ValueIter<'_>>
where
K: AsRef<[u8]>,
R: RangeBounds<K>,
{
let sb = range.start_bound();
let eb = range.end_bound();
self.0.range(self.1, get_bound(sb), get_bound(eb))
}
#[inline]
pub fn range_rev<K, R>(&self, range: R) -> Result<ValueIter<'_>>
where
K: AsRef<[u8]>,
R: RangeBounds<K>,
{
let sb = range.start_bound();
let eb = range.end_bound();
self.0.range_rev(self.1, get_bound(sb), get_bound(eb))
}
}
#[allow(clippy::len_without_is_empty)]
impl<'a> Transaction<'a> {
#[inline]
pub fn get<T: AsRef<[u8]>>(&self, tree: &Tree, key: T) -> TxOpResult<Option<Value>> {
self.0.get(tree.1, key.as_ref())
}
#[inline]
pub fn len(&self, tree: &Tree) -> TxOpResult<usize> {
self.0.len(tree.1)
}
/// Returns the old value if there was one
#[inline]
pub fn insert<T: AsRef<[u8]>, U: AsRef<[u8]>>(
&mut self,
tree: &Tree,
key: T,
value: U,
) -> TxOpResult<Option<Value>> {
self.0.insert(tree.1, key.as_ref(), value.as_ref())
}
/// Returns the old value if there was one
#[inline]
pub fn remove<T: AsRef<[u8]>>(&mut self, tree: &Tree, key: T) -> TxOpResult<Option<Value>> {
self.0.remove(tree.1, key.as_ref())
}
#[inline]
pub fn iter(&self, tree: &Tree) -> TxOpResult<TxValueIter<'_>> {
self.0.iter(tree.1)
}
#[inline]
pub fn iter_rev(&self, tree: &Tree) -> TxOpResult<TxValueIter<'_>> {
self.0.iter_rev(tree.1)
}
#[inline]
pub fn range<K, R>(&self, tree: &Tree, range: R) -> TxOpResult<TxValueIter<'_>>
where
K: AsRef<[u8]>,
R: RangeBounds<K>,
{
let sb = range.start_bound();
let eb = range.end_bound();
self.0.range(tree.1, get_bound(sb), get_bound(eb))
}
#[inline]
pub fn range_rev<K, R>(&self, tree: &Tree, range: R) -> TxOpResult<TxValueIter<'_>>
where
K: AsRef<[u8]>,
R: RangeBounds<K>,
{
let sb = range.start_bound();
let eb = range.end_bound();
self.0.range_rev(tree.1, get_bound(sb), get_bound(eb))
}
// ----
#[inline]
pub fn abort<R, E>(self, e: E) -> TxResult<R, E> {
Err(TxError::Abort(e))
}
#[inline]
pub fn commit<R, E>(self, r: R) -> TxResult<R, E> {
Ok(r)
}
}
// ---- Internal interfaces
pub(crate) trait IDb: Send + Sync {
fn engine(&self) -> String;
fn open_tree(&self, name: &str) -> Result<usize>;
fn list_trees(&self) -> Result<Vec<String>>;
fn get(&self, tree: usize, key: &[u8]) -> Result<Option<Value>>;
fn len(&self, tree: usize) -> Result<usize>;
fn insert(&self, tree: usize, key: &[u8], value: &[u8]) -> Result<Option<Value>>;
fn remove(&self, tree: usize, key: &[u8]) -> Result<Option<Value>>;
fn clear(&self, tree: usize) -> Result<()>;
fn iter(&self, tree: usize) -> Result<ValueIter<'_>>;
fn iter_rev(&self, tree: usize) -> Result<ValueIter<'_>>;
fn range<'r>(
&self,
tree: usize,
low: Bound<&'r [u8]>,
high: Bound<&'r [u8]>,
) -> Result<ValueIter<'_>>;
fn range_rev<'r>(
&self,
tree: usize,
low: Bound<&'r [u8]>,
high: Bound<&'r [u8]>,
) -> Result<ValueIter<'_>>;
fn transaction(&self, f: &dyn ITxFn) -> TxResult<(), ()>;
}
pub(crate) trait ITx {
fn get(&self, tree: usize, key: &[u8]) -> TxOpResult<Option<Value>>;
fn len(&self, tree: usize) -> TxOpResult<usize>;
fn insert(&mut self, tree: usize, key: &[u8], value: &[u8]) -> TxOpResult<Option<Value>>;
fn remove(&mut self, tree: usize, key: &[u8]) -> TxOpResult<Option<Value>>;
fn iter(&self, tree: usize) -> TxOpResult<TxValueIter<'_>>;
fn iter_rev(&self, tree: usize) -> TxOpResult<TxValueIter<'_>>;
fn range<'r>(
&self,
tree: usize,
low: Bound<&'r [u8]>,
high: Bound<&'r [u8]>,
) -> TxOpResult<TxValueIter<'_>>;
fn range_rev<'r>(
&self,
tree: usize,
low: Bound<&'r [u8]>,
high: Bound<&'r [u8]>,
) -> TxOpResult<TxValueIter<'_>>;
}
pub(crate) trait ITxFn {
fn try_on(&self, tx: &mut dyn ITx) -> TxFnResult;
}
pub(crate) enum TxFnResult {
Ok,
Abort,
DbErr,
}
struct TxFn<F, R, E>
where
F: Fn(Transaction<'_>) -> TxResult<R, E>,
{
function: F,
result: Cell<Option<TxResult<R, E>>>,
}
impl<F, R, E> ITxFn for TxFn<F, R, E>
where
F: Fn(Transaction<'_>) -> TxResult<R, E>,
{
fn try_on(&self, tx: &mut dyn ITx) -> TxFnResult {
let res = (self.function)(Transaction(tx));
let res2 = match &res {
Ok(_) => TxFnResult::Ok,
Err(TxError::Abort(_)) => TxFnResult::Abort,
Err(TxError::Db(_)) => TxFnResult::DbErr,
};
self.result.set(Some(res));
res2
}
}
// ----
fn get_bound<K: AsRef<[u8]>>(b: Bound<&K>) -> Bound<&[u8]> {
match b {
Bound::Included(v) => Bound::Included(v.as_ref()),
Bound::Excluded(v) => Bound::Excluded(v.as_ref()),
Bound::Unbounded => Bound::Unbounded,
}
}
+350
View File
@@ -0,0 +1,350 @@
use core::ops::Bound;
use core::ptr::NonNull;
use std::collections::HashMap;
use std::convert::TryInto;
use std::sync::{Arc, RwLock};
use heed::types::ByteSlice;
use heed::{BytesDecode, Env, RoTxn, RwTxn, UntypedDatabase as Database};
use crate::{
Db, Error, IDb, ITx, ITxFn, Result, TxError, TxFnResult, TxOpError, TxOpResult, TxResult,
TxValueIter, Value, ValueIter,
};
pub use heed;
// -- err
impl From<heed::Error> for Error {
fn from(e: heed::Error) -> Error {
Error(format!("LMDB: {}", e).into())
}
}
impl From<heed::Error> for TxOpError {
fn from(e: heed::Error) -> TxOpError {
TxOpError(e.into())
}
}
// -- db
pub struct LmdbDb {
db: heed::Env,
trees: RwLock<(Vec<Database>, HashMap<String, usize>)>,
}
impl LmdbDb {
pub fn init(db: Env) -> Db {
let s = Self {
db,
trees: RwLock::new((Vec::new(), HashMap::new())),
};
Db(Arc::new(s))
}
fn get_tree(&self, i: usize) -> Result<Database> {
self.trees
.read()
.unwrap()
.0
.get(i)
.cloned()
.ok_or_else(|| Error("invalid tree id".into()))
}
}
impl IDb for LmdbDb {
fn engine(&self) -> String {
"LMDB (using Heed crate)".into()
}
fn open_tree(&self, name: &str) -> Result<usize> {
let mut trees = self.trees.write().unwrap();
if let Some(i) = trees.1.get(name) {
Ok(*i)
} else {
let tree = self.db.create_database(Some(name))?;
let i = trees.0.len();
trees.0.push(tree);
trees.1.insert(name.to_string(), i);
Ok(i)
}
}
fn list_trees(&self) -> Result<Vec<String>> {
let tree0 = match self.db.open_database::<heed::types::Str, ByteSlice>(None)? {
Some(x) => x,
None => return Ok(vec![]),
};
let mut ret = vec![];
let tx = self.db.read_txn()?;
for item in tree0.iter(&tx)? {
let (tree_name, _) = item?;
ret.push(tree_name.to_string());
}
drop(tx);
let mut ret2 = vec![];
for tree_name in ret {
if self
.db
.open_database::<ByteSlice, ByteSlice>(Some(&tree_name))?
.is_some()
{
ret2.push(tree_name);
}
}
Ok(ret2)
}
// ----
fn get(&self, tree: usize, key: &[u8]) -> Result<Option<Value>> {
let tree = self.get_tree(tree)?;
let tx = self.db.read_txn()?;
let val = tree.get(&tx, key)?;
match val {
None => Ok(None),
Some(v) => Ok(Some(v.to_vec())),
}
}
fn len(&self, tree: usize) -> Result<usize> {
let tree = self.get_tree(tree)?;
let tx = self.db.read_txn()?;
Ok(tree.len(&tx)?.try_into().unwrap())
}
fn insert(&self, tree: usize, key: &[u8], value: &[u8]) -> Result<Option<Value>> {
let tree = self.get_tree(tree)?;
let mut tx = self.db.write_txn()?;
let old_val = tree.get(&tx, key)?.map(Vec::from);
tree.put(&mut tx, key, value)?;
tx.commit()?;
Ok(old_val)
}
fn remove(&self, tree: usize, key: &[u8]) -> Result<Option<Value>> {
let tree = self.get_tree(tree)?;
let mut tx = self.db.write_txn()?;
let old_val = tree.get(&tx, key)?.map(Vec::from);
tree.delete(&mut tx, key)?;
tx.commit()?;
Ok(old_val)
}
fn clear(&self, tree: usize) -> Result<()> {
let tree = self.get_tree(tree)?;
let mut tx = self.db.write_txn()?;
tree.clear(&mut tx)?;
tx.commit()?;
Ok(())
}
fn iter(&self, tree: usize) -> Result<ValueIter<'_>> {
let tree = self.get_tree(tree)?;
let tx = self.db.read_txn()?;
TxAndIterator::make(tx, |tx| Ok(tree.iter(tx)?))
}
fn iter_rev(&self, tree: usize) -> Result<ValueIter<'_>> {
let tree = self.get_tree(tree)?;
let tx = self.db.read_txn()?;
TxAndIterator::make(tx, |tx| Ok(tree.rev_iter(tx)?))
}
fn range<'r>(
&self,
tree: usize,
low: Bound<&'r [u8]>,
high: Bound<&'r [u8]>,
) -> Result<ValueIter<'_>> {
let tree = self.get_tree(tree)?;
let tx = self.db.read_txn()?;
TxAndIterator::make(tx, |tx| Ok(tree.range(tx, &(low, high))?))
}
fn range_rev<'r>(
&self,
tree: usize,
low: Bound<&'r [u8]>,
high: Bound<&'r [u8]>,
) -> Result<ValueIter<'_>> {
let tree = self.get_tree(tree)?;
let tx = self.db.read_txn()?;
TxAndIterator::make(tx, |tx| Ok(tree.rev_range(tx, &(low, high))?))
}
// ----
fn transaction(&self, f: &dyn ITxFn) -> TxResult<(), ()> {
let trees = self.trees.read().unwrap();
let mut tx = LmdbTx {
trees: &trees.0[..],
tx: self
.db
.write_txn()
.map_err(Error::from)
.map_err(TxError::Db)?,
};
let res = f.try_on(&mut tx);
match res {
TxFnResult::Ok => {
tx.tx.commit().map_err(Error::from).map_err(TxError::Db)?;
Ok(())
}
TxFnResult::Abort => {
tx.tx.abort().map_err(Error::from).map_err(TxError::Db)?;
Err(TxError::Abort(()))
}
TxFnResult::DbErr => {
tx.tx.abort().map_err(Error::from).map_err(TxError::Db)?;
Err(TxError::Db(Error(
"(this message will be discarded)".into(),
)))
}
}
}
}
// ----
struct LmdbTx<'a> {
trees: &'a [Database],
tx: RwTxn<'a, 'a>,
}
impl<'a> LmdbTx<'a> {
fn get_tree(&self, i: usize) -> TxOpResult<&Database> {
self.trees.get(i).ok_or_else(|| {
TxOpError(Error(
"invalid tree id (it might have been openned after the transaction started)".into(),
))
})
}
}
impl<'a> ITx for LmdbTx<'a> {
fn get(&self, tree: usize, key: &[u8]) -> TxOpResult<Option<Value>> {
let tree = self.get_tree(tree)?;
match tree.get(&self.tx, key)? {
Some(v) => Ok(Some(v.to_vec())),
None => Ok(None),
}
}
fn len(&self, _tree: usize) -> TxOpResult<usize> {
unimplemented!(".len() in transaction not supported with LMDB backend")
}
fn insert(&mut self, tree: usize, key: &[u8], value: &[u8]) -> TxOpResult<Option<Value>> {
let tree = *self.get_tree(tree)?;
let old_val = tree.get(&self.tx, key)?.map(Vec::from);
tree.put(&mut self.tx, key, value)?;
Ok(old_val)
}
fn remove(&mut self, tree: usize, key: &[u8]) -> TxOpResult<Option<Value>> {
let tree = *self.get_tree(tree)?;
let old_val = tree.get(&self.tx, key)?.map(Vec::from);
tree.delete(&mut self.tx, key)?;
Ok(old_val)
}
fn iter(&self, _tree: usize) -> TxOpResult<TxValueIter<'_>> {
unimplemented!("Iterators in transactions not supported with LMDB backend");
}
fn iter_rev(&self, _tree: usize) -> TxOpResult<TxValueIter<'_>> {
unimplemented!("Iterators in transactions not supported with LMDB backend");
}
fn range<'r>(
&self,
_tree: usize,
_low: Bound<&'r [u8]>,
_high: Bound<&'r [u8]>,
) -> TxOpResult<TxValueIter<'_>> {
unimplemented!("Iterators in transactions not supported with LMDB backend");
}
fn range_rev<'r>(
&self,
_tree: usize,
_low: Bound<&'r [u8]>,
_high: Bound<&'r [u8]>,
) -> TxOpResult<TxValueIter<'_>> {
unimplemented!("Iterators in transactions not supported with LMDB backend");
}
}
// ----
type IteratorItem<'a> = heed::Result<(
<ByteSlice as BytesDecode<'a>>::DItem,
<ByteSlice as BytesDecode<'a>>::DItem,
)>;
struct TxAndIterator<'a, I>
where
I: Iterator<Item = IteratorItem<'a>> + 'a,
{
tx: RoTxn<'a>,
iter: Option<I>,
}
impl<'a, I> TxAndIterator<'a, I>
where
I: Iterator<Item = IteratorItem<'a>> + 'a,
{
fn make<F>(tx: RoTxn<'a>, iterfun: F) -> Result<ValueIter<'a>>
where
F: FnOnce(&'a RoTxn<'a>) -> Result<I>,
{
let mut res = TxAndIterator { tx, iter: None };
let tx = unsafe { NonNull::from(&res.tx).as_ref() };
res.iter = Some(iterfun(tx)?);
Ok(Box::new(res))
}
}
impl<'a, I> Drop for TxAndIterator<'a, I>
where
I: Iterator<Item = IteratorItem<'a>> + 'a,
{
fn drop(&mut self) {
drop(self.iter.take());
}
}
impl<'a, I> Iterator for TxAndIterator<'a, I>
where
I: Iterator<Item = IteratorItem<'a>> + 'a,
{
type Item = Result<(Value, Value)>;
fn next(&mut self) -> Option<Self::Item> {
match self.iter.as_mut().unwrap().next() {
None => None,
Some(Err(e)) => Some(Err(e.into())),
Some(Ok((k, v))) => Some(Ok((k.to_vec(), v.to_vec()))),
}
}
}
// ----
#[cfg(target_pointer_width = "64")]
pub fn recommended_map_size() -> usize {
1usize << 40
}
#[cfg(target_pointer_width = "32")]
pub fn recommended_map_size() -> usize {
warn!("LMDB is not recommended on 32-bit systems, database size will be limited");
1usize << 30
}
+266
View File
@@ -0,0 +1,266 @@
use core::ops::Bound;
use std::cell::Cell;
use std::collections::HashMap;
use std::sync::{Arc, RwLock};
use sled::transaction::{
ConflictableTransactionError, TransactionError, Transactional, TransactionalTree,
UnabortableTransactionError,
};
use crate::{
Db, Error, IDb, ITx, ITxFn, Result, TxError, TxFnResult, TxOpError, TxOpResult, TxResult,
TxValueIter, Value, ValueIter,
};
pub use sled;
// -- err
impl From<sled::Error> for Error {
fn from(e: sled::Error) -> Error {
Error(format!("Sled: {}", e).into())
}
}
impl From<sled::Error> for TxOpError {
fn from(e: sled::Error) -> TxOpError {
TxOpError(e.into())
}
}
// -- db
pub struct SledDb {
db: sled::Db,
trees: RwLock<(Vec<sled::Tree>, HashMap<String, usize>)>,
}
impl SledDb {
pub fn init(db: sled::Db) -> Db {
let s = Self {
db,
trees: RwLock::new((Vec::new(), HashMap::new())),
};
Db(Arc::new(s))
}
fn get_tree(&self, i: usize) -> Result<sled::Tree> {
self.trees
.read()
.unwrap()
.0
.get(i)
.cloned()
.ok_or_else(|| Error("invalid tree id".into()))
}
}
impl IDb for SledDb {
fn engine(&self) -> String {
"Sled".into()
}
fn open_tree(&self, name: &str) -> Result<usize> {
let mut trees = self.trees.write().unwrap();
if let Some(i) = trees.1.get(name) {
Ok(*i)
} else {
let tree = self.db.open_tree(name)?;
let i = trees.0.len();
trees.0.push(tree);
trees.1.insert(name.to_string(), i);
Ok(i)
}
}
fn list_trees(&self) -> Result<Vec<String>> {
let mut trees = vec![];
for name in self.db.tree_names() {
let name = std::str::from_utf8(&name)
.map_err(|e| Error(format!("{}", e).into()))?
.to_string();
if name != "__sled__default" {
trees.push(name);
}
}
Ok(trees)
}
// ----
fn get(&self, tree: usize, key: &[u8]) -> Result<Option<Value>> {
let tree = self.get_tree(tree)?;
let val = tree.get(key)?;
Ok(val.map(|x| x.to_vec()))
}
fn len(&self, tree: usize) -> Result<usize> {
let tree = self.get_tree(tree)?;
Ok(tree.len())
}
fn insert(&self, tree: usize, key: &[u8], value: &[u8]) -> Result<Option<Value>> {
let tree = self.get_tree(tree)?;
let old_val = tree.insert(key, value)?;
Ok(old_val.map(|x| x.to_vec()))
}
fn remove(&self, tree: usize, key: &[u8]) -> Result<Option<Value>> {
let tree = self.get_tree(tree)?;
let old_val = tree.remove(key)?;
Ok(old_val.map(|x| x.to_vec()))
}
fn clear(&self, tree: usize) -> Result<()> {
let tree = self.get_tree(tree)?;
tree.clear()?;
Ok(())
}
fn iter(&self, tree: usize) -> Result<ValueIter<'_>> {
let tree = self.get_tree(tree)?;
Ok(Box::new(tree.iter().map(|v| {
v.map(|(x, y)| (x.to_vec(), y.to_vec())).map_err(Into::into)
})))
}
fn iter_rev(&self, tree: usize) -> Result<ValueIter<'_>> {
let tree = self.get_tree(tree)?;
Ok(Box::new(tree.iter().rev().map(|v| {
v.map(|(x, y)| (x.to_vec(), y.to_vec())).map_err(Into::into)
})))
}
fn range<'r>(
&self,
tree: usize,
low: Bound<&'r [u8]>,
high: Bound<&'r [u8]>,
) -> Result<ValueIter<'_>> {
let tree = self.get_tree(tree)?;
Ok(Box::new(tree.range::<&'r [u8], _>((low, high)).map(|v| {
v.map(|(x, y)| (x.to_vec(), y.to_vec())).map_err(Into::into)
})))
}
fn range_rev<'r>(
&self,
tree: usize,
low: Bound<&'r [u8]>,
high: Bound<&'r [u8]>,
) -> Result<ValueIter<'_>> {
let tree = self.get_tree(tree)?;
Ok(Box::new(tree.range::<&'r [u8], _>((low, high)).rev().map(
|v| v.map(|(x, y)| (x.to_vec(), y.to_vec())).map_err(Into::into),
)))
}
// ----
fn transaction(&self, f: &dyn ITxFn) -> TxResult<(), ()> {
let trees = self.trees.read().unwrap();
let res = trees.0.transaction(|txtrees| {
let mut tx = SledTx {
trees: txtrees,
err: Cell::new(None),
};
match f.try_on(&mut tx) {
TxFnResult::Ok => {
assert!(tx.err.into_inner().is_none());
Ok(())
}
TxFnResult::Abort => {
assert!(tx.err.into_inner().is_none());
Err(ConflictableTransactionError::Abort(()))
}
TxFnResult::DbErr => {
let e = tx.err.into_inner().expect("No DB error");
Err(e.into())
}
}
});
match res {
Ok(()) => Ok(()),
Err(TransactionError::Abort(())) => Err(TxError::Abort(())),
Err(TransactionError::Storage(s)) => Err(TxError::Db(s.into())),
}
}
}
// ----
struct SledTx<'a> {
trees: &'a [TransactionalTree],
err: Cell<Option<UnabortableTransactionError>>,
}
impl<'a> SledTx<'a> {
fn get_tree(&self, i: usize) -> TxOpResult<&TransactionalTree> {
self.trees.get(i).ok_or_else(|| {
TxOpError(Error(
"invalid tree id (it might have been openned after the transaction started)".into(),
))
})
}
fn save_error<R>(
&self,
v: std::result::Result<R, UnabortableTransactionError>,
) -> TxOpResult<R> {
match v {
Ok(x) => Ok(x),
Err(e) => {
let txt = format!("{}", e);
self.err.set(Some(e));
Err(TxOpError(Error(txt.into())))
}
}
}
}
impl<'a> ITx for SledTx<'a> {
fn get(&self, tree: usize, key: &[u8]) -> TxOpResult<Option<Value>> {
let tree = self.get_tree(tree)?;
let tmp = self.save_error(tree.get(key))?;
Ok(tmp.map(|x| x.to_vec()))
}
fn len(&self, _tree: usize) -> TxOpResult<usize> {
unimplemented!(".len() in transaction not supported with Sled backend")
}
fn insert(&mut self, tree: usize, key: &[u8], value: &[u8]) -> TxOpResult<Option<Value>> {
let tree = self.get_tree(tree)?;
let old_val = self.save_error(tree.insert(key, value))?;
Ok(old_val.map(|x| x.to_vec()))
}
fn remove(&mut self, tree: usize, key: &[u8]) -> TxOpResult<Option<Value>> {
let tree = self.get_tree(tree)?;
let old_val = self.save_error(tree.remove(key))?;
Ok(old_val.map(|x| x.to_vec()))
}
fn iter(&self, _tree: usize) -> TxOpResult<TxValueIter<'_>> {
unimplemented!("Iterators in transactions not supported with Sled backend");
}
fn iter_rev(&self, _tree: usize) -> TxOpResult<TxValueIter<'_>> {
unimplemented!("Iterators in transactions not supported with Sled backend");
}
fn range<'r>(
&self,
_tree: usize,
_low: Bound<&'r [u8]>,
_high: Bound<&'r [u8]>,
) -> TxOpResult<TxValueIter<'_>> {
unimplemented!("Iterators in transactions not supported with Sled backend");
}
fn range_rev<'r>(
&self,
_tree: usize,
_low: Bound<&'r [u8]>,
_high: Bound<&'r [u8]>,
) -> TxOpResult<TxValueIter<'_>> {
unimplemented!("Iterators in transactions not supported with Sled backend");
}
}
+508
View File
@@ -0,0 +1,508 @@
use core::ops::Bound;
use std::borrow::BorrowMut;
use std::marker::PhantomPinned;
use std::pin::Pin;
use std::ptr::NonNull;
use std::sync::{Arc, Mutex, MutexGuard};
use rusqlite::{params, Connection, Rows, Statement, Transaction};
use crate::{
Db, Error, IDb, ITx, ITxFn, Result, TxError, TxFnResult, TxOpError, TxOpResult, TxResult,
TxValueIter, Value, ValueIter,
};
pub use rusqlite;
// --- err
impl From<rusqlite::Error> for Error {
fn from(e: rusqlite::Error) -> Error {
Error(format!("Sqlite: {}", e).into())
}
}
impl From<rusqlite::Error> for TxOpError {
fn from(e: rusqlite::Error) -> TxOpError {
TxOpError(e.into())
}
}
// -- db
pub struct SqliteDb(Mutex<SqliteDbInner>);
struct SqliteDbInner {
db: Connection,
trees: Vec<String>,
}
impl SqliteDb {
pub fn init(db: rusqlite::Connection) -> Db {
let s = Self(Mutex::new(SqliteDbInner {
db,
trees: Vec::new(),
}));
Db(Arc::new(s))
}
}
impl SqliteDbInner {
fn get_tree(&self, i: usize) -> Result<&'_ str> {
self.trees
.get(i)
.map(String::as_str)
.ok_or_else(|| Error("invalid tree id".into()))
}
fn internal_get(&self, tree: &str, key: &[u8]) -> Result<Option<Value>> {
let mut stmt = self
.db
.prepare(&format!("SELECT v FROM {} WHERE k = ?1", tree))?;
let mut res_iter = stmt.query([key])?;
match res_iter.next()? {
None => Ok(None),
Some(v) => Ok(Some(v.get::<_, Vec<u8>>(0)?)),
}
}
}
impl IDb for SqliteDb {
fn engine(&self) -> String {
format!("sqlite3 v{} (using rusqlite crate)", rusqlite::version())
}
fn open_tree(&self, name: &str) -> Result<usize> {
let name = format!("tree_{}", name.replace(':', "_COLON_"));
let mut this = self.0.lock().unwrap();
if let Some(i) = this.trees.iter().position(|x| x == &name) {
Ok(i)
} else {
trace!("create table {}", name);
this.db.execute(
&format!(
"CREATE TABLE IF NOT EXISTS {} (
k BLOB PRIMARY KEY,
v BLOB
)",
name
),
[],
)?;
trace!("table created: {}, unlocking", name);
let i = this.trees.len();
this.trees.push(name.to_string());
Ok(i)
}
}
fn list_trees(&self) -> Result<Vec<String>> {
let mut trees = vec![];
trace!("list_trees: lock db");
let this = self.0.lock().unwrap();
trace!("list_trees: lock acquired");
let mut stmt = this.db.prepare(
"SELECT name FROM sqlite_schema WHERE type = 'table' AND name LIKE 'tree_%'",
)?;
let mut rows = stmt.query([])?;
while let Some(row) = rows.next()? {
let name = row.get::<_, String>(0)?;
let name = name.replace("_COLON_", ":");
let name = name.strip_prefix("tree_").unwrap().to_string();
trees.push(name);
}
Ok(trees)
}
// ----
fn get(&self, tree: usize, key: &[u8]) -> Result<Option<Value>> {
trace!("get {}: lock db", tree);
let this = self.0.lock().unwrap();
trace!("get {}: lock acquired", tree);
let tree = this.get_tree(tree)?;
this.internal_get(tree, key)
}
fn len(&self, tree: usize) -> Result<usize> {
trace!("len {}: lock db", tree);
let this = self.0.lock().unwrap();
trace!("len {}: lock acquired", tree);
let tree = this.get_tree(tree)?;
let mut stmt = this.db.prepare(&format!("SELECT COUNT(*) FROM {}", tree))?;
let mut res_iter = stmt.query([])?;
match res_iter.next()? {
None => Ok(0),
Some(v) => Ok(v.get::<_, usize>(0)?),
}
}
fn insert(&self, tree: usize, key: &[u8], value: &[u8]) -> Result<Option<Value>> {
trace!("insert {}: lock db", tree);
let this = self.0.lock().unwrap();
trace!("insert {}: lock acquired", tree);
let tree = this.get_tree(tree)?;
let old_val = this.internal_get(tree, key)?;
let sql = match &old_val {
Some(_) => format!("UPDATE {} SET v = ?2 WHERE k = ?1", tree),
None => format!("INSERT INTO {} (k, v) VALUES (?1, ?2)", tree),
};
let n = this.db.execute(&sql, params![key, value])?;
assert_eq!(n, 1);
Ok(old_val)
}
fn remove(&self, tree: usize, key: &[u8]) -> Result<Option<Value>> {
trace!("remove {}: lock db", tree);
let this = self.0.lock().unwrap();
trace!("remove {}: lock acquired", tree);
let tree = this.get_tree(tree)?;
let old_val = this.internal_get(tree, key)?;
if old_val.is_some() {
let n = this
.db
.execute(&format!("DELETE FROM {} WHERE k = ?1", tree), params![key])?;
assert_eq!(n, 1);
}
Ok(old_val)
}
fn clear(&self, tree: usize) -> Result<()> {
trace!("clear {}: lock db", tree);
let this = self.0.lock().unwrap();
trace!("clear {}: lock acquired", tree);
let tree = this.get_tree(tree)?;
this.db.execute(&format!("DELETE FROM {}", tree), [])?;
Ok(())
}
fn iter(&self, tree: usize) -> Result<ValueIter<'_>> {
trace!("iter {}: lock db", tree);
let this = self.0.lock().unwrap();
trace!("iter {}: lock acquired", tree);
let tree = this.get_tree(tree)?;
let sql = format!("SELECT k, v FROM {} ORDER BY k ASC", tree);
DbValueIterator::make(this, &sql, [])
}
fn iter_rev(&self, tree: usize) -> Result<ValueIter<'_>> {
trace!("iter_rev {}: lock db", tree);
let this = self.0.lock().unwrap();
trace!("iter_rev {}: lock acquired", tree);
let tree = this.get_tree(tree)?;
let sql = format!("SELECT k, v FROM {} ORDER BY k DESC", tree);
DbValueIterator::make(this, &sql, [])
}
fn range<'r>(
&self,
tree: usize,
low: Bound<&'r [u8]>,
high: Bound<&'r [u8]>,
) -> Result<ValueIter<'_>> {
trace!("range {}: lock db", tree);
let this = self.0.lock().unwrap();
trace!("range {}: lock acquired", tree);
let tree = this.get_tree(tree)?;
let (bounds_sql, params) = bounds_sql(low, high);
let sql = format!("SELECT k, v FROM {} {} ORDER BY k ASC", tree, bounds_sql);
let params = params
.iter()
.map(|x| x as &dyn rusqlite::ToSql)
.collect::<Vec<_>>();
DbValueIterator::make::<&[&dyn rusqlite::ToSql]>(this, &sql, params.as_ref())
}
fn range_rev<'r>(
&self,
tree: usize,
low: Bound<&'r [u8]>,
high: Bound<&'r [u8]>,
) -> Result<ValueIter<'_>> {
trace!("range_rev {}: lock db", tree);
let this = self.0.lock().unwrap();
trace!("range_rev {}: lock acquired", tree);
let tree = this.get_tree(tree)?;
let (bounds_sql, params) = bounds_sql(low, high);
let sql = format!("SELECT k, v FROM {} {} ORDER BY k DESC", tree, bounds_sql);
let params = params
.iter()
.map(|x| x as &dyn rusqlite::ToSql)
.collect::<Vec<_>>();
DbValueIterator::make::<&[&dyn rusqlite::ToSql]>(this, &sql, params.as_ref())
}
// ----
fn transaction(&self, f: &dyn ITxFn) -> TxResult<(), ()> {
trace!("transaction: lock db");
let mut this = self.0.lock().unwrap();
trace!("transaction: lock acquired");
let this_mut_ref: &mut SqliteDbInner = this.borrow_mut();
let mut tx = SqliteTx {
tx: this_mut_ref
.db
.transaction()
.map_err(Error::from)
.map_err(TxError::Db)?,
trees: &this_mut_ref.trees,
};
let res = match f.try_on(&mut tx) {
TxFnResult::Ok => {
tx.tx.commit().map_err(Error::from).map_err(TxError::Db)?;
Ok(())
}
TxFnResult::Abort => {
tx.tx.rollback().map_err(Error::from).map_err(TxError::Db)?;
Err(TxError::Abort(()))
}
TxFnResult::DbErr => {
tx.tx.rollback().map_err(Error::from).map_err(TxError::Db)?;
Err(TxError::Db(Error(
"(this message will be discarded)".into(),
)))
}
};
trace!("transaction done");
res
}
}
// ----
struct SqliteTx<'a> {
tx: Transaction<'a>,
trees: &'a [String],
}
impl<'a> SqliteTx<'a> {
fn get_tree(&self, i: usize) -> TxOpResult<&'_ str> {
self.trees.get(i).map(String::as_ref).ok_or_else(|| {
TxOpError(Error(
"invalid tree id (it might have been openned after the transaction started)".into(),
))
})
}
fn internal_get(&self, tree: &str, key: &[u8]) -> TxOpResult<Option<Value>> {
let mut stmt = self
.tx
.prepare(&format!("SELECT v FROM {} WHERE k = ?1", tree))?;
let mut res_iter = stmt.query([key])?;
match res_iter.next()? {
None => Ok(None),
Some(v) => Ok(Some(v.get::<_, Vec<u8>>(0)?)),
}
}
}
impl<'a> ITx for SqliteTx<'a> {
fn get(&self, tree: usize, key: &[u8]) -> TxOpResult<Option<Value>> {
let tree = self.get_tree(tree)?;
self.internal_get(tree, key)
}
fn len(&self, tree: usize) -> TxOpResult<usize> {
let tree = self.get_tree(tree)?;
let mut stmt = self.tx.prepare(&format!("SELECT COUNT(*) FROM {}", tree))?;
let mut res_iter = stmt.query([])?;
match res_iter.next()? {
None => Ok(0),
Some(v) => Ok(v.get::<_, usize>(0)?),
}
}
fn insert(&mut self, tree: usize, key: &[u8], value: &[u8]) -> TxOpResult<Option<Value>> {
let tree = self.get_tree(tree)?;
let old_val = self.internal_get(tree, key)?;
let sql = match &old_val {
Some(_) => format!("UPDATE {} SET v = ?2 WHERE k = ?1", tree),
None => format!("INSERT INTO {} (k, v) VALUES (?1, ?2)", tree),
};
let n = self.tx.execute(&sql, params![key, value])?;
assert_eq!(n, 1);
Ok(old_val)
}
fn remove(&mut self, tree: usize, key: &[u8]) -> TxOpResult<Option<Value>> {
let tree = self.get_tree(tree)?;
let old_val = self.internal_get(tree, key)?;
if old_val.is_some() {
let n = self
.tx
.execute(&format!("DELETE FROM {} WHERE k = ?1", tree), params![key])?;
assert_eq!(n, 1);
}
Ok(old_val)
}
fn iter(&self, _tree: usize) -> TxOpResult<TxValueIter<'_>> {
unimplemented!();
}
fn iter_rev(&self, _tree: usize) -> TxOpResult<TxValueIter<'_>> {
unimplemented!();
}
fn range<'r>(
&self,
_tree: usize,
_low: Bound<&'r [u8]>,
_high: Bound<&'r [u8]>,
) -> TxOpResult<TxValueIter<'_>> {
unimplemented!();
}
fn range_rev<'r>(
&self,
_tree: usize,
_low: Bound<&'r [u8]>,
_high: Bound<&'r [u8]>,
) -> TxOpResult<TxValueIter<'_>> {
unimplemented!();
}
}
// ----
struct DbValueIterator<'a> {
db: MutexGuard<'a, SqliteDbInner>,
stmt: Option<Statement<'a>>,
iter: Option<Rows<'a>>,
_pin: PhantomPinned,
}
impl<'a> DbValueIterator<'a> {
fn make<P: rusqlite::Params>(
db: MutexGuard<'a, SqliteDbInner>,
sql: &str,
args: P,
) -> Result<ValueIter<'a>> {
let res = DbValueIterator {
db,
stmt: None,
iter: None,
_pin: PhantomPinned,
};
let mut boxed = Box::pin(res);
trace!("make iterator with sql: {}", sql);
unsafe {
let db = NonNull::from(&boxed.db);
let stmt = db.as_ref().db.prepare(sql)?;
let mut_ref: Pin<&mut DbValueIterator<'a>> = Pin::as_mut(&mut boxed);
Pin::get_unchecked_mut(mut_ref).stmt = Some(stmt);
let mut stmt = NonNull::from(&boxed.stmt);
let iter = stmt.as_mut().as_mut().unwrap().query(args)?;
let mut_ref: Pin<&mut DbValueIterator<'a>> = Pin::as_mut(&mut boxed);
Pin::get_unchecked_mut(mut_ref).iter = Some(iter);
}
Ok(Box::new(DbValueIteratorPin(boxed)))
}
}
impl<'a> Drop for DbValueIterator<'a> {
fn drop(&mut self) {
trace!("drop iter");
drop(self.iter.take());
drop(self.stmt.take());
}
}
struct DbValueIteratorPin<'a>(Pin<Box<DbValueIterator<'a>>>);
impl<'a> Iterator for DbValueIteratorPin<'a> {
type Item = Result<(Value, Value)>;
fn next(&mut self) -> Option<Self::Item> {
let next = unsafe {
let mut_ref: Pin<&mut DbValueIterator<'a>> = Pin::as_mut(&mut self.0);
Pin::get_unchecked_mut(mut_ref).iter.as_mut()?.next()
};
let row = match next {
Err(e) => return Some(Err(e.into())),
Ok(None) => return None,
Ok(Some(r)) => r,
};
let k = match row.get::<_, Vec<u8>>(0) {
Err(e) => return Some(Err(e.into())),
Ok(x) => x,
};
let v = match row.get::<_, Vec<u8>>(1) {
Err(e) => return Some(Err(e.into())),
Ok(y) => y,
};
Some(Ok((k, v)))
}
}
// ----
fn bounds_sql<'r>(low: Bound<&'r [u8]>, high: Bound<&'r [u8]>) -> (String, Vec<Vec<u8>>) {
let mut sql = String::new();
let mut params: Vec<Vec<u8>> = vec![];
match low {
Bound::Included(b) => {
sql.push_str(" WHERE k >= ?1");
params.push(b.to_vec());
}
Bound::Excluded(b) => {
sql.push_str(" WHERE k > ?1");
params.push(b.to_vec());
}
Bound::Unbounded => (),
};
match high {
Bound::Included(b) => {
if !params.is_empty() {
sql.push_str(" AND k <= ?2");
} else {
sql.push_str(" WHERE k <= ?1");
}
params.push(b.to_vec());
}
Bound::Excluded(b) => {
if !params.is_empty() {
sql.push_str(" AND k < ?2");
} else {
sql.push_str(" WHERE k < ?1");
}
params.push(b.to_vec());
}
Bound::Unbounded => (),
}
(sql, params)
}
+106
View File
@@ -0,0 +1,106 @@
use crate::*;
use crate::lmdb_adapter::LmdbDb;
use crate::sled_adapter::SledDb;
use crate::sqlite_adapter::SqliteDb;
fn test_suite(db: Db) {
let tree = db.open_tree("tree").unwrap();
let ka: &[u8] = &b"test"[..];
let kb: &[u8] = &b"zwello"[..];
let kint: &[u8] = &b"tz"[..];
let va: &[u8] = &b"plop"[..];
let vb: &[u8] = &b"plip"[..];
let vc: &[u8] = &b"plup"[..];
assert!(tree.insert(ka, va).unwrap().is_none());
assert_eq!(tree.get(ka).unwrap().unwrap(), va);
let res = db.transaction::<_, (), _>(|mut tx| {
assert_eq!(tx.get(&tree, ka).unwrap().unwrap(), va);
assert_eq!(tx.insert(&tree, ka, vb).unwrap().unwrap(), va);
assert_eq!(tx.get(&tree, ka).unwrap().unwrap(), vb);
tx.commit(12)
});
assert!(matches!(res, Ok(12)));
assert_eq!(tree.get(ka).unwrap().unwrap(), vb);
let res = db.transaction::<(), _, _>(|mut tx| {
assert_eq!(tx.get(&tree, ka).unwrap().unwrap(), vb);
assert_eq!(tx.insert(&tree, ka, vc).unwrap().unwrap(), vb);
assert_eq!(tx.get(&tree, ka).unwrap().unwrap(), vc);
tx.abort(42)
});
assert!(matches!(res, Err(TxError::Abort(42))));
assert_eq!(tree.get(ka).unwrap().unwrap(), vb);
let mut iter = tree.iter().unwrap();
let next = iter.next().unwrap().unwrap();
assert_eq!((next.0.as_ref(), next.1.as_ref()), (ka, vb));
assert!(iter.next().is_none());
drop(iter);
assert!(tree.insert(kb, vc).unwrap().is_none());
assert_eq!(tree.get(kb).unwrap().unwrap(), vc);
let mut iter = tree.iter().unwrap();
let next = iter.next().unwrap().unwrap();
assert_eq!((next.0.as_ref(), next.1.as_ref()), (ka, vb));
let next = iter.next().unwrap().unwrap();
assert_eq!((next.0.as_ref(), next.1.as_ref()), (kb, vc));
assert!(iter.next().is_none());
drop(iter);
let mut iter = tree.range(kint..).unwrap();
let next = iter.next().unwrap().unwrap();
assert_eq!((next.0.as_ref(), next.1.as_ref()), (kb, vc));
assert!(iter.next().is_none());
drop(iter);
let mut iter = tree.range_rev(..kint).unwrap();
let next = iter.next().unwrap().unwrap();
assert_eq!((next.0.as_ref(), next.1.as_ref()), (ka, vb));
assert!(iter.next().is_none());
drop(iter);
let mut iter = tree.iter_rev().unwrap();
let next = iter.next().unwrap().unwrap();
assert_eq!((next.0.as_ref(), next.1.as_ref()), (kb, vc));
let next = iter.next().unwrap().unwrap();
assert_eq!((next.0.as_ref(), next.1.as_ref()), (ka, vb));
assert!(iter.next().is_none());
drop(iter);
}
#[test]
fn test_lmdb_db() {
let path = mktemp::Temp::new_dir().unwrap();
let db = heed::EnvOpenOptions::new()
.max_dbs(100)
.open(&path)
.unwrap();
let db = LmdbDb::init(db);
test_suite(db);
drop(path);
}
#[test]
fn test_sled_db() {
let path = mktemp::Temp::new_dir().unwrap();
let db = SledDb::init(sled::open(path.to_path_buf()).unwrap());
test_suite(db);
drop(path);
}
#[test]
fn test_sqlite_db() {
let db = SqliteDb::init(rusqlite::Connection::open_in_memory().unwrap());
test_suite(db);
}
+49 -17
View File
@@ -1,6 +1,6 @@
[package]
name = "garage"
version = "0.7.0"
version = "0.8.0"
authors = ["Alex Auvolat <alex@adnab.me>"]
edition = "2018"
license = "AGPL-3.0"
@@ -21,25 +21,25 @@ path = "tests/lib.rs"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
garage_api = { version = "0.7.0", path = "../api" }
garage_model = { version = "0.7.0", path = "../model" }
garage_rpc = { version = "0.7.0", path = "../rpc" }
garage_table = { version = "0.7.0", path = "../table" }
garage_util = { version = "0.7.0", path = "../util" }
garage_web = { version = "0.7.0", path = "../web" }
garage_admin = { version = "0.7.0", path = "../admin" }
garage_db = { version = "0.8.0", path = "../db" }
garage_api = { version = "0.8.0", path = "../api" }
garage_block = { version = "0.8.0", path = "../block" }
garage_model = { version = "0.8.0", path = "../model" }
garage_rpc = { version = "0.8.0", path = "../rpc" }
garage_table = { version = "0.8.0", path = "../table" }
garage_util = { version = "0.8.0", path = "../util" }
garage_web = { version = "0.8.0", path = "../web" }
bytes = "1.0"
git-version = "0.3.4"
bytesize = "1.1"
timeago = "0.3"
hex = "0.4"
tracing = { version = "0.1.30", features = ["log-always"] }
pretty_env_logger = "0.4"
tracing-subscriber = { version = "0.3", features = ["env-filter"] }
rand = "0.8"
async-trait = "0.1.7"
sodiumoxide = { version = "0.2.5-0", package = "kuska-sodiumoxide" }
sled = "0.34"
rmp-serde = "0.15"
serde = { version = "1.0", default-features = false, features = ["derive", "rc"] }
serde_bytes = "0.11"
@@ -50,16 +50,48 @@ futures = "0.3"
futures-util = "0.3"
tokio = { version = "1.0", default-features = false, features = ["rt", "rt-multi-thread", "io-util", "net", "time", "macros", "sync", "signal", "fs"] }
#netapp = { version = "0.3.2", git = "https://git.deuxfleurs.fr/lx/netapp" }
#netapp = { version = "0.4", path = "../../../netapp" }
netapp = "0.4"
netapp = "0.5"
opentelemetry = { version = "0.17", features = [ "rt-tokio" ] }
opentelemetry-prometheus = { version = "0.10", optional = true }
opentelemetry-otlp = { version = "0.10", optional = true }
prometheus = { version = "0.13", optional = true }
[dev-dependencies]
aws-sdk-s3 = "0.8"
chrono = "0.4"
http = "0.2"
hmac = "0.10"
hmac = "0.12"
hyper = { version = "0.14", features = ["client", "http1", "runtime"] }
sha2 = "0.9"
sha2 = "0.10"
static_init = "1.0"
assert-json-diff = "2.0"
serde_json = "1.0"
base64 = "0.13"
[features]
default = [ "bundled-libs", "metrics", "sled" ]
k2v = [ "garage_util/k2v", "garage_api/k2v" ]
# Database engines, Sled is still our default even though we don't like it
sled = [ "garage_model/sled" ]
lmdb = [ "garage_model/lmdb" ]
sqlite = [ "garage_model/sqlite" ]
# Automatic registration and discovery via Kubernetes API
kubernetes-discovery = [ "garage_rpc/kubernetes-discovery" ]
# Prometheus exporter (/metrics endpoint).
metrics = [ "garage_api/metrics", "opentelemetry-prometheus", "prometheus" ]
# Exporter for the OpenTelemetry Collector.
telemetry-otlp = [ "opentelemetry-otlp" ]
# NOTE: bundled-libs and system-libs should be treat as mutually exclusive;
# exactly one of them should be enabled.
# Use bundled libsqlite instead of linking against system-provided.
bundled-libs = [ "garage_db/bundled-libs" ]
# Link against system-provided libsodium and libzstd.
system-libs = [ "garage_block/system-libs", "garage_rpc/system-libs", "sodiumoxide/use-pkg-config" ]
+189 -80
View File
@@ -15,34 +15,45 @@ use garage_table::*;
use garage_rpc::*;
use garage_block::repair::ScrubWorkerCommand;
use garage_model::bucket_alias_table::*;
use garage_model::bucket_table::*;
use garage_model::garage::Garage;
use garage_model::helper::error::{Error, OkOrBadRequest};
use garage_model::key_table::*;
use garage_model::migrate::Migrate;
use garage_model::object_table::ObjectFilter;
use garage_model::permission::*;
use crate::cli::*;
use crate::repair::Repair;
use crate::repair::online::launch_online_repair;
pub const ADMIN_RPC_PATH: &str = "garage/admin_rpc.rs/Rpc";
#[derive(Debug, Serialize, Deserialize)]
#[allow(clippy::large_enum_variant)]
pub enum AdminRpc {
BucketOperation(BucketOperation),
KeyOperation(KeyOperation),
LaunchRepair(RepairOpt),
Migrate(MigrateOpt),
Stats(StatsOpt),
Worker(WorkerOpt),
// Replies
Ok(String),
BucketList(Vec<Bucket>),
BucketInfo(Bucket, HashMap<String, Key>),
BucketInfo {
bucket: Bucket,
relevant_keys: HashMap<String, Key>,
counters: HashMap<String, i64>,
},
KeyList(Vec<(String, String)>),
KeyInfo(Key, HashMap<Uuid, Bucket>),
WorkerList(
HashMap<usize, garage_util::background::WorkerInfo>,
WorkerListOpt,
),
}
impl Rpc for AdminRpc {
@@ -73,6 +84,7 @@ impl AdminRpcHandler {
BucketOperation::Allow(query) => self.handle_bucket_allow(query).await,
BucketOperation::Deny(query) => self.handle_bucket_deny(query).await,
BucketOperation::Website(query) => self.handle_bucket_website(query).await,
BucketOperation::SetQuotas(query) => self.handle_bucket_set_quotas(query).await,
}
}
@@ -80,8 +92,15 @@ impl AdminRpcHandler {
let buckets = self
.garage
.bucket_table
.get_range(&EmptyKey, None, Some(DeletedFilter::NotDeleted), 10000)
.get_range(
&EmptyKey,
None,
Some(DeletedFilter::NotDeleted),
10000,
EnumerationOrder::Forward,
)
.await?;
Ok(AdminRpc::BucketList(buckets))
}
@@ -99,6 +118,15 @@ impl AdminRpcHandler {
.get_existing_bucket(bucket_id)
.await?;
let counters = self
.garage
.object_counter_table
.table
.get(&bucket_id, &EmptyKey)
.await?
.map(|x| x.filtered_values(&self.garage.system.ring.borrow()))
.unwrap_or_default();
let mut relevant_keys = HashMap::new();
for (k, _) in bucket
.state
@@ -134,7 +162,11 @@ impl AdminRpcHandler {
}
}
Ok(AdminRpc::BucketInfo(bucket, relevant_keys))
Ok(AdminRpc::BucketInfo {
bucket,
relevant_keys,
counters,
})
}
#[allow(clippy::ptr_arg)]
@@ -207,12 +239,7 @@ impl AdminRpcHandler {
}
// Check bucket is empty
let objects = self
.garage
.object_table
.get_range(&bucket_id, None, Some(ObjectFilter::IsData), 10)
.await?;
if !objects.is_empty() {
if !helper.is_bucket_empty(bucket_id).await? {
return Err(Error::BadRequest(format!(
"Bucket {} is not empty",
query.name
@@ -249,6 +276,7 @@ impl AdminRpcHandler {
async fn handle_alias_bucket(&self, query: &AliasBucketOpt) -> Result<AdminRpc, Error> {
let helper = self.garage.bucket_helper();
let key_helper = self.garage.key_helper();
let bucket_id = helper
.resolve_global_bucket_name(&query.existing_bucket)
@@ -256,7 +284,7 @@ impl AdminRpcHandler {
.ok_or_bad_request("Bucket not found")?;
if let Some(key_pattern) = &query.local {
let key = helper.get_existing_matching_key(key_pattern).await?;
let key = key_helper.get_existing_matching_key(key_pattern).await?;
helper
.set_local_bucket_alias(bucket_id, &key.key_id, &query.new_name)
@@ -278,9 +306,10 @@ impl AdminRpcHandler {
async fn handle_unalias_bucket(&self, query: &UnaliasBucketOpt) -> Result<AdminRpc, Error> {
let helper = self.garage.bucket_helper();
let key_helper = self.garage.key_helper();
if let Some(key_pattern) = &query.local {
let key = helper.get_existing_matching_key(key_pattern).await?;
let key = key_helper.get_existing_matching_key(key_pattern).await?;
let bucket_id = key
.state
@@ -319,12 +348,15 @@ impl AdminRpcHandler {
async fn handle_bucket_allow(&self, query: &PermBucketOpt) -> Result<AdminRpc, Error> {
let helper = self.garage.bucket_helper();
let key_helper = self.garage.key_helper();
let bucket_id = helper
.resolve_global_bucket_name(&query.bucket)
.await?
.ok_or_bad_request("Bucket not found")?;
let key = helper.get_existing_matching_key(&query.key_pattern).await?;
let key = key_helper
.get_existing_matching_key(&query.key_pattern)
.await?;
let allow_read = query.read || key.allow_read(&bucket_id);
let allow_write = query.write || key.allow_write(&bucket_id);
@@ -351,12 +383,15 @@ impl AdminRpcHandler {
async fn handle_bucket_deny(&self, query: &PermBucketOpt) -> Result<AdminRpc, Error> {
let helper = self.garage.bucket_helper();
let key_helper = self.garage.key_helper();
let bucket_id = helper
.resolve_global_bucket_name(&query.bucket)
.await?
.ok_or_bad_request("Bucket not found")?;
let key = helper.get_existing_matching_key(&query.key_pattern).await?;
let key = key_helper
.get_existing_matching_key(&query.key_pattern)
.await?;
let allow_read = !query.read && key.allow_read(&bucket_id);
let allow_write = !query.write && key.allow_write(&bucket_id);
@@ -423,6 +458,60 @@ impl AdminRpcHandler {
Ok(AdminRpc::Ok(msg))
}
async fn handle_bucket_set_quotas(&self, query: &SetQuotasOpt) -> Result<AdminRpc, Error> {
let bucket_id = self
.garage
.bucket_helper()
.resolve_global_bucket_name(&query.bucket)
.await?
.ok_or_bad_request("Bucket not found")?;
let mut bucket = self
.garage
.bucket_helper()
.get_existing_bucket(bucket_id)
.await?;
let bucket_state = bucket.state.as_option_mut().unwrap();
if query.max_size.is_none() && query.max_objects.is_none() {
return Err(Error::BadRequest(
"You must specify either --max-size or --max-objects (or both) for this command to do something.".to_string(),
));
}
let mut quotas = bucket_state.quotas.get().clone();
match query.max_size.as_ref().map(String::as_ref) {
Some("none") => quotas.max_size = None,
Some(v) => {
let bs = v
.parse::<bytesize::ByteSize>()
.ok_or_bad_request(format!("Invalid size specified: {}", v))?;
quotas.max_size = Some(bs.as_u64());
}
_ => (),
}
match query.max_objects.as_ref().map(String::as_ref) {
Some("none") => quotas.max_objects = None,
Some(v) => {
let mo = v
.parse::<u64>()
.ok_or_bad_request(format!("Invalid number specified: {}", v))?;
quotas.max_objects = Some(mo);
}
_ => (),
}
bucket_state.quotas.update(quotas);
self.garage.bucket_table.insert(&bucket).await?;
Ok(AdminRpc::Ok(format!(
"Quotas updated for {}",
&query.bucket
)))
}
async fn handle_key_cmd(&self, cmd: &KeyOperation) -> Result<AdminRpc, Error> {
match cmd {
KeyOperation::List => self.handle_list_keys().await,
@@ -445,6 +534,7 @@ impl AdminRpcHandler {
None,
Some(KeyFilter::Deleted(DeletedFilter::NotDeleted)),
10000,
EnumerationOrder::Forward,
)
.await?
.iter()
@@ -456,7 +546,7 @@ impl AdminRpcHandler {
async fn handle_key_info(&self, query: &KeyOpt) -> Result<AdminRpc, Error> {
let key = self
.garage
.bucket_helper()
.key_helper()
.get_existing_matching_key(&query.key_pattern)
.await?;
self.key_info_result(key).await
@@ -471,7 +561,7 @@ impl AdminRpcHandler {
async fn handle_rename_key(&self, query: &KeyRenameOpt) -> Result<AdminRpc, Error> {
let mut key = self
.garage
.bucket_helper()
.key_helper()
.get_existing_matching_key(&query.key_pattern)
.await?;
key.params_mut()
@@ -483,9 +573,11 @@ impl AdminRpcHandler {
}
async fn handle_delete_key(&self, query: &KeyDeleteOpt) -> Result<AdminRpc, Error> {
let helper = self.garage.bucket_helper();
let key_helper = self.garage.key_helper();
let mut key = helper.get_existing_matching_key(&query.key_pattern).await?;
let mut key = key_helper
.get_existing_matching_key(&query.key_pattern)
.await?;
if !query.yes {
return Err(Error::BadRequest(
@@ -493,32 +585,7 @@ impl AdminRpcHandler {
));
}
let state = key.state.as_option_mut().unwrap();
// --- done checking, now commit ---
// (the step at unset_local_bucket_alias will fail if a bucket
// does not have another alias, the deletion will be
// interrupted in the middle if that happens)
// 1. Delete local aliases
for (alias, _, to) in state.local_aliases.items().iter() {
if let Some(bucket_id) = to {
helper
.unset_local_bucket_alias(*bucket_id, &key.key_id, alias)
.await?;
}
}
// 2. Remove permissions on all authorized buckets
for (ab_id, _auth) in state.authorized_buckets.items().iter() {
helper
.set_bucket_key_permissions(*ab_id, &key.key_id, BucketKeyPerm::NO_PERMISSIONS)
.await?;
}
// 3. Actually delete key
key.state = Deletable::delete();
self.garage.key_table.insert(&key).await?;
key_helper.delete_key(&mut key).await?;
Ok(AdminRpc::Ok(format!(
"Key {} was deleted successfully.",
@@ -529,7 +596,7 @@ impl AdminRpcHandler {
async fn handle_allow_key(&self, query: &KeyPermOpt) -> Result<AdminRpc, Error> {
let mut key = self
.garage
.bucket_helper()
.key_helper()
.get_existing_matching_key(&query.key_pattern)
.await?;
if query.create_bucket {
@@ -542,7 +609,7 @@ impl AdminRpcHandler {
async fn handle_deny_key(&self, query: &KeyPermOpt) -> Result<AdminRpc, Error> {
let mut key = self
.garage
.bucket_helper()
.key_helper()
.get_existing_matching_key(&query.key_pattern)
.await?;
if query.create_bucket {
@@ -616,7 +683,7 @@ impl AdminRpcHandler {
.endpoint
.call(
&node,
&AdminRpc::LaunchRepair(opt_to_send.clone()),
AdminRpc::LaunchRepair(opt_to_send.clone()),
PRIO_NORMAL,
)
.await;
@@ -633,15 +700,7 @@ impl AdminRpcHandler {
)))
}
} else {
let repair = Repair {
garage: self.garage.clone(),
};
self.garage
.system
.background
.spawn_worker("Repair worker".into(), move |must_exit| async move {
repair.repair_worker(opt, must_exit).await
});
launch_online_repair(self.garage.clone(), opt).await;
Ok(AdminRpc::Ok(format!(
"Repair launched on {:?}",
self.garage.system.id
@@ -664,7 +723,7 @@ impl AdminRpcHandler {
let node_id = (*node).into();
match self
.endpoint
.call(&node_id, &AdminRpc::Stats(opt), PRIO_NORMAL)
.call(&node_id, AdminRpc::Stats(opt), PRIO_NORMAL)
.await?
{
Ok(AdminRpc::Ok(s)) => writeln!(&mut ret, "{}", s).unwrap(),
@@ -674,22 +733,22 @@ impl AdminRpcHandler {
}
Ok(AdminRpc::Ok(ret))
} else {
Ok(AdminRpc::Ok(self.gather_stats_local(opt)))
Ok(AdminRpc::Ok(self.gather_stats_local(opt)?))
}
}
fn gather_stats_local(&self, opt: StatsOpt) -> String {
fn gather_stats_local(&self, opt: StatsOpt) -> Result<String, Error> {
let mut ret = String::new();
writeln!(
&mut ret,
"\nGarage version: {}",
option_env!("GIT_VERSION").unwrap_or(git_version::git_version!(
prefix = "git:",
cargo_prefix = "cargo:",
fallback = "unknown"
))
"\nGarage version: {} [features: {}]",
garage_util::version::garage_version(),
garage_util::version::garage_features()
.map(|list| list.join(", "))
.unwrap_or_else(|| "(unknown)".into()),
)
.unwrap();
writeln!(&mut ret, "\nDatabase engine: {}", self.garage.db.engine()).unwrap();
// Gather ring statistics
let ring = self.garage.system.ring.borrow().clone();
@@ -707,59 +766,108 @@ impl AdminRpcHandler {
writeln!(&mut ret, " {:?} {}", n, c).unwrap();
}
self.gather_table_stats(&mut ret, &self.garage.bucket_table, &opt);
self.gather_table_stats(&mut ret, &self.garage.key_table, &opt);
self.gather_table_stats(&mut ret, &self.garage.object_table, &opt);
self.gather_table_stats(&mut ret, &self.garage.version_table, &opt);
self.gather_table_stats(&mut ret, &self.garage.block_ref_table, &opt);
self.gather_table_stats(&mut ret, &self.garage.bucket_table, &opt)?;
self.gather_table_stats(&mut ret, &self.garage.key_table, &opt)?;
self.gather_table_stats(&mut ret, &self.garage.object_table, &opt)?;
self.gather_table_stats(&mut ret, &self.garage.version_table, &opt)?;
self.gather_table_stats(&mut ret, &self.garage.block_ref_table, &opt)?;
writeln!(&mut ret, "\nBlock manager stats:").unwrap();
if opt.detailed {
writeln!(
&mut ret,
" number of RC entries (~= number of blocks): {}",
self.garage.block_manager.rc_len()
self.garage.block_manager.rc_len()?
)
.unwrap();
}
writeln!(
&mut ret,
" resync queue length: {}",
self.garage.block_manager.resync_queue_len()
self.garage.block_manager.resync.queue_len()?
)
.unwrap();
writeln!(
&mut ret,
" blocks with resync errors: {}",
self.garage.block_manager.resync_errors_len()
self.garage.block_manager.resync.errors_len()?
)
.unwrap();
ret
Ok(ret)
}
fn gather_table_stats<F, R>(&self, to: &mut String, t: &Arc<Table<F, R>>, opt: &StatsOpt)
fn gather_table_stats<F, R>(
&self,
to: &mut String,
t: &Arc<Table<F, R>>,
opt: &StatsOpt,
) -> Result<(), Error>
where
F: TableSchema + 'static,
R: TableReplication + 'static,
{
writeln!(to, "\nTable stats for {}", F::TABLE_NAME).unwrap();
if opt.detailed {
writeln!(to, " number of items: {}", t.data.store.len()).unwrap();
writeln!(
to,
" number of items: {}",
t.data.store.len().map_err(GarageError::from)?
)
.unwrap();
writeln!(
to,
" Merkle tree size: {}",
t.merkle_updater.merkle_tree_len()
t.merkle_updater.merkle_tree_len()?
)
.unwrap();
}
writeln!(
to,
" Merkle updater todo queue length: {}",
t.merkle_updater.todo_len()
t.merkle_updater.todo_len()?
)
.unwrap();
writeln!(to, " GC todo queue length: {}", t.data.gc_todo_len()).unwrap();
writeln!(to, " GC todo queue length: {}", t.data.gc_todo_len()?).unwrap();
Ok(())
}
// ----
async fn handle_worker_cmd(&self, opt: WorkerOpt) -> Result<AdminRpc, Error> {
match opt.cmd {
WorkerCmd::List { opt } => {
let workers = self.garage.background.get_worker_info();
Ok(AdminRpc::WorkerList(workers, opt))
}
WorkerCmd::Set { opt } => match opt {
WorkerSetCmd::ScrubTranquility { tranquility } => {
let scrub_command = ScrubWorkerCommand::SetTranquility(tranquility);
self.garage
.block_manager
.send_scrub_command(scrub_command)
.await;
Ok(AdminRpc::Ok("Scrub tranquility updated".into()))
}
WorkerSetCmd::ResyncNWorkers { n_workers } => {
self.garage
.block_manager
.resync
.set_n_workers(n_workers)
.await?;
Ok(AdminRpc::Ok("Number of resync workers updated".into()))
}
WorkerSetCmd::ResyncTranquility { tranquility } => {
self.garage
.block_manager
.resync
.set_tranquility(tranquility)
.await?;
Ok(AdminRpc::Ok("Resync tranquility updated".into()))
}
},
}
}
}
@@ -776,6 +884,7 @@ impl EndpointHandler<AdminRpc> for AdminRpcHandler {
AdminRpc::Migrate(opt) => self.handle_migrate(opt.clone()).await,
AdminRpc::LaunchRepair(opt) => self.handle_launch_repair(opt.clone()).await,
AdminRpc::Stats(opt) => self.handle_stats(opt.clone()).await,
AdminRpc::Worker(opt) => self.handle_worker_cmd(opt.clone()).await,
m => Err(GarageError::unexpected_rpc_message(m).into()),
}
}
+21 -9
View File
@@ -1,6 +1,8 @@
use std::collections::HashSet;
use std::time::Duration;
use garage_util::error::*;
use garage_util::formater::format_table;
use garage_rpc::layout::*;
use garage_rpc::system::*;
@@ -38,13 +40,14 @@ pub async fn cli_command_dispatch(
cmd_admin(admin_rpc_endpoint, rpc_host, AdminRpc::LaunchRepair(ro)).await
}
Command::Stats(so) => cmd_admin(admin_rpc_endpoint, rpc_host, AdminRpc::Stats(so)).await,
Command::Worker(wo) => cmd_admin(admin_rpc_endpoint, rpc_host, AdminRpc::Worker(wo)).await,
_ => unreachable!(),
}
}
pub async fn cmd_status(rpc_cli: &Endpoint<SystemRpc, ()>, rpc_host: NodeID) -> Result<(), Error> {
let status = match rpc_cli
.call(&rpc_host, &SystemRpc::GetKnownNodes, PRIO_NORMAL)
.call(&rpc_host, SystemRpc::GetKnownNodes, PRIO_NORMAL)
.await??
{
SystemRpc::ReturnKnownNodes(nodes) => nodes,
@@ -85,19 +88,21 @@ pub async fn cmd_status(rpc_cli: &Endpoint<SystemRpc, ()>, rpc_host: NodeID) ->
format_table(healthy_nodes);
let status_keys = status.iter().map(|adv| adv.id).collect::<HashSet<_>>();
let failure_case_1 = status.iter().any(|adv| !adv.is_up);
let failure_case_1 = status
.iter()
.any(|adv| !adv.is_up && matches!(layout.roles.get(&adv.id), Some(NodeRoleV(Some(_)))));
let failure_case_2 = layout
.roles
.items()
.iter()
.filter(|(_, _, v)| v.0.is_some())
.any(|(id, _, _)| !status_keys.contains(id));
.any(|(id, _, v)| !status_keys.contains(id) && v.0.is_some());
if failure_case_1 || failure_case_2 {
println!("\n==== FAILED NODES ====");
let mut failed_nodes =
vec!["ID\tHostname\tAddress\tTags\tZone\tCapacity\tLast seen".to_string()];
for adv in status.iter().filter(|adv| !adv.is_up) {
if let Some(NodeRoleV(Some(cfg))) = layout.roles.get(&adv.id) {
let tf = timeago::Formatter::new();
failed_nodes.push(format!(
"{id:?}\t{host}\t{addr}\t[{tags}]\t{zone}\t{capacity}\t{last_seen}",
id = adv.id,
@@ -108,7 +113,7 @@ pub async fn cmd_status(rpc_cli: &Endpoint<SystemRpc, ()>, rpc_host: NodeID) ->
capacity = cfg.capacity_string(),
last_seen = adv
.last_seen_secs_ago
.map(|s| format!("{}s ago", s))
.map(|s| tf.convert(Duration::from_secs(s)))
.unwrap_or_else(|| "never seen".into()),
));
}
@@ -144,7 +149,7 @@ pub async fn cmd_connect(
args: ConnectNodeOpt,
) -> Result<(), Error> {
match rpc_cli
.call(&rpc_host, &SystemRpc::Connect(args.node), PRIO_NORMAL)
.call(&rpc_host, SystemRpc::Connect(args.node), PRIO_NORMAL)
.await??
{
SystemRpc::Ok => {
@@ -160,15 +165,19 @@ pub async fn cmd_admin(
rpc_host: NodeID,
args: AdminRpc,
) -> Result<(), HelperError> {
match rpc_cli.call(&rpc_host, &args, PRIO_NORMAL).await?? {
match rpc_cli.call(&rpc_host, args, PRIO_NORMAL).await?? {
AdminRpc::Ok(msg) => {
println!("{}", msg);
}
AdminRpc::BucketList(bl) => {
print_bucket_list(bl);
}
AdminRpc::BucketInfo(bucket, rk) => {
print_bucket_info(&bucket, &rk);
AdminRpc::BucketInfo {
bucket,
relevant_keys,
counters,
} => {
print_bucket_info(&bucket, &relevant_keys, &counters);
}
AdminRpc::KeyList(kl) => {
print_key_list(kl);
@@ -176,6 +185,9 @@ pub async fn cmd_admin(
AdminRpc::KeyInfo(key, rb) => {
print_key_info(&key, &rb);
}
AdminRpc::WorkerList(wi, wlo) => {
print_worker_info(wi, wlo);
}
r => {
error!("Unexpected response: {:?}", r);
}
+29 -50
View File
@@ -1,6 +1,6 @@
use garage_util::crdt::Crdt;
use garage_util::data::*;
use garage_util::error::*;
use garage_util::formater::format_table;
use garage_rpc::layout::*;
use garage_rpc::system::*;
@@ -36,21 +36,29 @@ pub async fn cmd_assign_role(
args: AssignRoleOpt,
) -> Result<(), Error> {
let status = match rpc_cli
.call(&rpc_host, &SystemRpc::GetKnownNodes, PRIO_NORMAL)
.call(&rpc_host, SystemRpc::GetKnownNodes, PRIO_NORMAL)
.await??
{
SystemRpc::ReturnKnownNodes(nodes) => nodes,
resp => return Err(Error::Message(format!("Invalid RPC response: {:?}", resp))),
};
let mut layout = fetch_layout(rpc_cli, rpc_host).await?;
let added_nodes = args
.node_ids
.iter()
.map(|node_id| find_matching_node(status.iter().map(|adv| adv.id), node_id))
.map(|node_id| {
find_matching_node(
status
.iter()
.map(|adv| adv.id)
.chain(layout.node_ids().iter().cloned()),
node_id,
)
})
.collect::<Result<Vec<_>, _>>()?;
let mut layout = fetch_layout(rpc_cli, rpc_host).await?;
let mut roles = layout.roles.clone();
roles.merge(&layout.staging);
@@ -203,31 +211,9 @@ pub async fn cmd_apply_layout(
rpc_host: NodeID,
apply_opt: ApplyLayoutOpt,
) -> Result<(), Error> {
let mut layout = fetch_layout(rpc_cli, rpc_host).await?;
let layout = fetch_layout(rpc_cli, rpc_host).await?;
match apply_opt.version {
None => {
println!("Please pass the --version flag to ensure that you are writing the correct version of the cluster layout.");
println!("To know the correct value of the --version flag, invoke `garage layout show` and review the proposed changes.");
return Err(Error::Message("--version flag is missing".into()));
}
Some(v) => {
if v != layout.version + 1 {
return Err(Error::Message("Invalid value of --version flag".into()));
}
}
}
layout.roles.merge(&layout.staging);
if !layout.calculate_partition_assignation() {
return Err(Error::Message("Could not calculate new assignation of partitions to nodes. This can happen if there are less nodes than the desired number of copies of your data (see the replication_mode configuration parameter).".into()));
}
layout.staging.clear();
layout.staging_hash = blake2sum(&rmp_to_vec_all_named(&layout.staging).unwrap()[..]);
layout.version += 1;
let layout = layout.apply_staged_changes(apply_opt.version)?;
send_layout(rpc_cli, rpc_host, layout).await?;
@@ -242,25 +228,9 @@ pub async fn cmd_revert_layout(
rpc_host: NodeID,
revert_opt: RevertLayoutOpt,
) -> Result<(), Error> {
let mut layout = fetch_layout(rpc_cli, rpc_host).await?;
let layout = fetch_layout(rpc_cli, rpc_host).await?;
match revert_opt.version {
None => {
println!("Please pass the --version flag to ensure that you are writing the correct version of the cluster layout.");
println!("To know the correct value of the --version flag, invoke `garage layout show` and review the proposed changes.");
return Err(Error::Message("--version flag is missing".into()));
}
Some(v) => {
if v != layout.version + 1 {
return Err(Error::Message("Invalid value of --version flag".into()));
}
}
}
layout.staging.clear();
layout.staging_hash = blake2sum(&rmp_to_vec_all_named(&layout.staging).unwrap()[..]);
layout.version += 1;
let layout = layout.revert_staged_changes(revert_opt.version)?;
send_layout(rpc_cli, rpc_host, layout).await?;
@@ -275,7 +245,7 @@ pub async fn fetch_layout(
rpc_host: NodeID,
) -> Result<ClusterLayout, Error> {
match rpc_cli
.call(&rpc_host, &SystemRpc::PullClusterLayout, PRIO_NORMAL)
.call(&rpc_host, SystemRpc::PullClusterLayout, PRIO_NORMAL)
.await??
{
SystemRpc::AdvertiseClusterLayout(t) => Ok(t),
@@ -291,7 +261,7 @@ pub async fn send_layout(
rpc_cli
.call(
&rpc_host,
&SystemRpc::AdvertiseClusterLayout(layout),
SystemRpc::AdvertiseClusterLayout(layout),
PRIO_NORMAL,
)
.await??;
@@ -323,11 +293,20 @@ pub fn print_cluster_layout(layout: &ClusterLayout) -> bool {
}
pub fn print_staging_role_changes(layout: &ClusterLayout) -> bool {
if !layout.staging.items().is_empty() {
let has_changes = layout
.staging
.items()
.iter()
.any(|(k, _, v)| layout.roles.get(k) != Some(v));
if has_changes {
println!();
println!("==== STAGED ROLE CHANGES ====");
let mut table = vec!["ID\tTags\tZone\tCapacity".to_string()];
for (id, _, role) in layout.staging.items().iter() {
if layout.roles.get(id) == Some(role) {
continue;
}
if let Some(role) = &role.0 {
let tags = role.tags.join(",");
table.push(format!(
+159 -43
View File
@@ -1,55 +1,65 @@
use serde::{Deserialize, Serialize};
use structopt::StructOpt;
use garage_util::version::garage_version;
#[derive(StructOpt, Debug)]
pub enum Command {
/// Run Garage server
#[structopt(name = "server")]
#[structopt(name = "server", version = garage_version())]
Server,
/// Get network status
#[structopt(name = "status")]
#[structopt(name = "status", version = garage_version())]
Status,
/// Operations on individual Garage nodes
#[structopt(name = "node")]
#[structopt(name = "node", version = garage_version())]
Node(NodeOperation),
/// Operations on the assignation of node roles in the cluster layout
#[structopt(name = "layout")]
#[structopt(name = "layout", version = garage_version())]
Layout(LayoutOperation),
/// Operations on buckets
#[structopt(name = "bucket")]
#[structopt(name = "bucket", version = garage_version())]
Bucket(BucketOperation),
/// Operations on S3 access keys
#[structopt(name = "key")]
#[structopt(name = "key", version = garage_version())]
Key(KeyOperation),
/// Run migrations from previous Garage version
/// (DO NOT USE WITHOUT READING FULL DOCUMENTATION)
#[structopt(name = "migrate")]
#[structopt(name = "migrate", version = garage_version())]
Migrate(MigrateOpt),
/// Start repair of node data
#[structopt(name = "repair")]
/// Start repair of node data on remote node
#[structopt(name = "repair", version = garage_version())]
Repair(RepairOpt),
/// Offline reparation of node data (these repairs must be run offline
/// directly on the server node)
#[structopt(name = "offline-repair", version = garage_version())]
OfflineRepair(OfflineRepairOpt),
/// Gather node statistics
#[structopt(name = "stats")]
#[structopt(name = "stats", version = garage_version())]
Stats(StatsOpt),
/// Manage background workers
#[structopt(name = "worker", version = garage_version())]
Worker(WorkerOpt),
}
#[derive(StructOpt, Debug)]
pub enum NodeOperation {
/// Print identifier (public key) of this Garage node
#[structopt(name = "id")]
#[structopt(name = "id", version = garage_version())]
NodeId(NodeIdOpt),
/// Connect to Garage node that is currently isolated from the system
#[structopt(name = "connect")]
#[structopt(name = "connect", version = garage_version())]
Connect(ConnectNodeOpt),
}
@@ -70,23 +80,23 @@ pub struct ConnectNodeOpt {
#[derive(StructOpt, Debug)]
pub enum LayoutOperation {
/// Assign role to Garage node
#[structopt(name = "assign")]
#[structopt(name = "assign", version = garage_version())]
Assign(AssignRoleOpt),
/// Remove role from Garage cluster node
#[structopt(name = "remove")]
#[structopt(name = "remove", version = garage_version())]
Remove(RemoveRoleOpt),
/// Show roles currently assigned to nodes and changes staged for commit
#[structopt(name = "show")]
#[structopt(name = "show", version = garage_version())]
Show,
/// Apply staged changes to cluster layout
#[structopt(name = "apply")]
#[structopt(name = "apply", version = garage_version())]
Apply(ApplyLayoutOpt),
/// Revert staged changes to cluster layout
#[structopt(name = "revert")]
#[structopt(name = "revert", version = garage_version())]
Revert(RevertLayoutOpt),
}
@@ -141,40 +151,44 @@ pub struct RevertLayoutOpt {
#[derive(Serialize, Deserialize, StructOpt, Debug)]
pub enum BucketOperation {
/// List buckets
#[structopt(name = "list")]
#[structopt(name = "list", version = garage_version())]
List,
/// Get bucket info
#[structopt(name = "info")]
#[structopt(name = "info", version = garage_version())]
Info(BucketOpt),
/// Create bucket
#[structopt(name = "create")]
#[structopt(name = "create", version = garage_version())]
Create(BucketOpt),
/// Delete bucket
#[structopt(name = "delete")]
#[structopt(name = "delete", version = garage_version())]
Delete(DeleteBucketOpt),
/// Alias bucket under new name
#[structopt(name = "alias")]
#[structopt(name = "alias", version = garage_version())]
Alias(AliasBucketOpt),
/// Remove bucket alias
#[structopt(name = "unalias")]
#[structopt(name = "unalias", version = garage_version())]
Unalias(UnaliasBucketOpt),
/// Allow key to read or write to bucket
#[structopt(name = "allow")]
#[structopt(name = "allow", version = garage_version())]
Allow(PermBucketOpt),
/// Deny key from reading or writing to bucket
#[structopt(name = "deny")]
#[structopt(name = "deny", version = garage_version())]
Deny(PermBucketOpt),
/// Expose as website or not
#[structopt(name = "website")]
#[structopt(name = "website", version = garage_version())]
Website(WebsiteOpt),
/// Set the quotas for this bucket
#[structopt(name = "set-quotas", version = garage_version())]
SetQuotas(SetQuotasOpt),
}
#[derive(Serialize, Deserialize, StructOpt, Debug)]
@@ -261,38 +275,53 @@ pub struct PermBucketOpt {
pub bucket: String,
}
#[derive(Serialize, Deserialize, StructOpt, Debug)]
pub struct SetQuotasOpt {
/// Bucket name
pub bucket: String,
/// Set a maximum size for the bucket (specify a size e.g. in MiB or GiB,
/// or `none` for no size restriction)
#[structopt(long = "max-size")]
pub max_size: Option<String>,
/// Set a maximum number of objects for the bucket (or `none` for no restriction)
#[structopt(long = "max-objects")]
pub max_objects: Option<String>,
}
#[derive(Serialize, Deserialize, StructOpt, Debug)]
pub enum KeyOperation {
/// List keys
#[structopt(name = "list")]
#[structopt(name = "list", version = garage_version())]
List,
/// Get key info
#[structopt(name = "info")]
#[structopt(name = "info", version = garage_version())]
Info(KeyOpt),
/// Create new key
#[structopt(name = "new")]
#[structopt(name = "new", version = garage_version())]
New(KeyNewOpt),
/// Rename key
#[structopt(name = "rename")]
#[structopt(name = "rename", version = garage_version())]
Rename(KeyRenameOpt),
/// Delete key
#[structopt(name = "delete")]
#[structopt(name = "delete", version = garage_version())]
Delete(KeyDeleteOpt),
/// Set permission flags for key
#[structopt(name = "allow")]
#[structopt(name = "allow", version = garage_version())]
Allow(KeyPermOpt),
/// Unset permission flags for key
#[structopt(name = "deny")]
#[structopt(name = "deny", version = garage_version())]
Deny(KeyPermOpt),
/// Import key
#[structopt(name = "import")]
#[structopt(name = "import", version = garage_version())]
Import(KeyImportOpt),
}
@@ -364,7 +393,7 @@ pub struct MigrateOpt {
#[derive(Serialize, Deserialize, StructOpt, Debug, Eq, PartialEq, Clone)]
pub enum MigrateWhat {
/// Migrate buckets and permissions from v0.5.0
#[structopt(name = "buckets050")]
#[structopt(name = "buckets050", version = garage_version())]
Buckets050,
}
@@ -385,26 +414,68 @@ pub struct RepairOpt {
#[derive(Serialize, Deserialize, StructOpt, Debug, Eq, PartialEq, Clone)]
pub enum RepairWhat {
/// Only do a full sync of metadata tables
#[structopt(name = "tables")]
#[structopt(name = "tables", version = garage_version())]
Tables,
/// Only repair (resync/rebalance) the set of stored blocks
#[structopt(name = "blocks")]
#[structopt(name = "blocks", version = garage_version())]
Blocks,
/// Only redo the propagation of object deletions to the version table (slow)
#[structopt(name = "versions")]
#[structopt(name = "versions", version = garage_version())]
Versions,
/// Only redo the propagation of version deletions to the block ref table (extremely slow)
#[structopt(name = "block_refs")]
#[structopt(name = "block_refs", version = garage_version())]
BlockRefs,
/// Verify integrity of all blocks on disc (extremely slow, i/o intensive)
#[structopt(name = "scrub")]
#[structopt(name = "scrub", version = garage_version())]
Scrub {
/// Tranquility factor (see tranquilizer documentation)
#[structopt(name = "tranquility", default_value = "2")]
#[structopt(subcommand)]
cmd: ScrubCmd,
},
}
#[derive(Serialize, Deserialize, StructOpt, Debug, Eq, PartialEq, Clone)]
pub enum ScrubCmd {
/// Start scrub
#[structopt(name = "start", version = garage_version())]
Start,
/// Pause scrub (it will resume automatically after 24 hours)
#[structopt(name = "pause", version = garage_version())]
Pause,
/// Resume paused scrub
#[structopt(name = "resume", version = garage_version())]
Resume,
/// Cancel scrub in progress
#[structopt(name = "cancel", version = garage_version())]
Cancel,
/// Set tranquility level for in-progress and future scrubs
#[structopt(name = "set-tranquility", version = garage_version())]
SetTranquility {
#[structopt()]
tranquility: u32,
},
}
#[derive(Serialize, Deserialize, StructOpt, Debug, Clone)]
pub struct OfflineRepairOpt {
/// Confirm the launch of the repair operation
#[structopt(long = "yes")]
pub yes: bool,
#[structopt(subcommand)]
pub what: OfflineRepairWhat,
}
#[derive(Serialize, Deserialize, StructOpt, Debug, Eq, PartialEq, Clone)]
pub enum OfflineRepairWhat {
/// Repair K2V item counters
#[cfg(feature = "k2v")]
#[structopt(name = "k2v_item_counters", version = garage_version())]
K2VItemCounters,
/// Repair object counters
#[structopt(name = "object_counters", version = garage_version())]
ObjectCounters,
}
#[derive(Serialize, Deserialize, StructOpt, Debug, Clone)]
pub struct StatsOpt {
/// Gather statistics from all nodes
@@ -415,3 +486,48 @@ pub struct StatsOpt {
#[structopt(short = "d", long = "detailed")]
pub detailed: bool,
}
#[derive(Serialize, Deserialize, StructOpt, Debug, Clone)]
pub struct WorkerOpt {
#[structopt(subcommand)]
pub cmd: WorkerCmd,
}
#[derive(Serialize, Deserialize, StructOpt, Debug, Eq, PartialEq, Clone)]
pub enum WorkerCmd {
/// List all workers on Garage node
#[structopt(name = "list", version = garage_version())]
List {
#[structopt(flatten)]
opt: WorkerListOpt,
},
/// Set worker parameter
#[structopt(name = "set", version = garage_version())]
Set {
#[structopt(subcommand)]
opt: WorkerSetCmd,
},
}
#[derive(Serialize, Deserialize, StructOpt, Debug, Eq, PartialEq, Clone, Copy)]
pub struct WorkerListOpt {
/// Show only busy workers
#[structopt(short = "b", long = "busy")]
pub busy: bool,
/// Show only workers with errors
#[structopt(short = "e", long = "errors")]
pub errors: bool,
}
#[derive(Serialize, Deserialize, StructOpt, Debug, Eq, PartialEq, Clone)]
pub enum WorkerSetCmd {
/// Set tranquility of scrub operations
#[structopt(name = "scrub-tranquility", version = garage_version())]
ScrubTranquility { tranquility: u32 },
/// Set number of concurrent block resync workers
#[structopt(name = "resync-n-workers", version = garage_version())]
ResyncNWorkers { n_workers: usize },
/// Set tranquility of block resync operations
#[structopt(name = "resync-tranquility", version = garage_version())]
ResyncTranquility { tranquility: u32 },
}
+104 -33
View File
@@ -1,11 +1,18 @@
use std::collections::HashMap;
use std::time::Duration;
use garage_util::background::*;
use garage_util::crdt::*;
use garage_util::data::Uuid;
use garage_util::error::*;
use garage_util::formater::format_table;
use garage_util::time::*;
use garage_model::bucket_table::*;
use garage_model::key_table::*;
use garage_model::s3::object_table::{BYTES, OBJECTS, UNFINISHED_UPLOADS};
use crate::cli::structs::WorkerListOpt;
pub fn print_bucket_list(bl: Vec<Bucket>) {
println!("List of buckets:");
@@ -28,11 +35,12 @@ pub fn print_bucket_list(bl: Vec<Bucket>) {
[((k, n), _, _)] => format!("{}:{}", k, n),
s => format!("[{} local aliases]", s.len()),
};
table.push(format!(
"\t{}\t{}\t{}",
aliases.join(","),
local_aliases_n,
hex::encode(bucket.id)
hex::encode(bucket.id),
));
}
format_table(table);
@@ -120,7 +128,11 @@ pub fn print_key_info(key: &Key, relevant_buckets: &HashMap<Uuid, Bucket>) {
}
}
pub fn print_bucket_info(bucket: &Bucket, relevant_keys: &HashMap<String, Key>) {
pub fn print_bucket_info(
bucket: &Bucket,
relevant_keys: &HashMap<String, Key>,
counters: &HashMap<String, i64>,
) {
let key_name = |k| {
relevant_keys
.get(k)
@@ -132,7 +144,42 @@ pub fn print_bucket_info(bucket: &Bucket, relevant_keys: &HashMap<String, Key>)
match &bucket.state {
Deletable::Deleted => println!("Bucket is deleted."),
Deletable::Present(p) => {
println!("Website access: {}", p.website_config.get().is_some());
let size =
bytesize::ByteSize::b(counters.get(BYTES).cloned().unwrap_or_default() as u64);
println!(
"\nSize: {} ({})",
size.to_string_as(true),
size.to_string_as(false)
);
println!(
"Objects: {}",
counters.get(OBJECTS).cloned().unwrap_or_default()
);
println!(
"Unfinished multipart uploads: {}",
counters
.get(UNFINISHED_UPLOADS)
.cloned()
.unwrap_or_default()
);
println!("\nWebsite access: {}", p.website_config.get().is_some());
let quotas = p.quotas.get();
if quotas.max_size.is_some() || quotas.max_objects.is_some() {
println!("\nQuotas:");
if let Some(ms) = quotas.max_size {
let ms = bytesize::ByteSize::b(ms);
println!(
" maximum size: {} ({})",
ms.to_string_as(true),
ms.to_string_as(false)
);
}
if let Some(mo) = quotas.max_objects {
println!(" maximum number of objects: {}", mo);
}
}
println!("\nGlobal aliases:");
for (alias, _, active) in p.aliases.items().iter() {
@@ -173,42 +220,13 @@ pub fn print_bucket_info(bucket: &Bucket, relevant_keys: &HashMap<String, Key>)
};
}
pub fn format_table(data: Vec<String>) {
let data = data
.iter()
.map(|s| s.split('\t').collect::<Vec<_>>())
.collect::<Vec<_>>();
let columns = data.iter().map(|row| row.len()).fold(0, std::cmp::max);
let mut column_size = vec![0; columns];
let mut out = String::new();
for row in data.iter() {
for (i, col) in row.iter().enumerate() {
column_size[i] = std::cmp::max(column_size[i], col.chars().count());
}
}
for row in data.iter() {
for (col, col_len) in row[..row.len() - 1].iter().zip(column_size.iter()) {
out.push_str(col);
(0..col_len - col.chars().count() + 2).for_each(|_| out.push(' '));
}
out.push_str(row[row.len() - 1]);
out.push('\n');
}
print!("{}", out);
}
pub fn find_matching_node(
cand: impl std::iter::Iterator<Item = Uuid>,
pattern: &str,
) -> Result<Uuid, Error> {
let mut candidates = vec![];
for c in cand {
if hex::encode(&c).starts_with(&pattern) {
if hex::encode(&c).starts_with(&pattern) && !candidates.contains(&c) {
candidates.push(c);
}
}
@@ -222,3 +240,56 @@ pub fn find_matching_node(
Ok(candidates[0])
}
}
pub fn print_worker_info(wi: HashMap<usize, WorkerInfo>, wlo: WorkerListOpt) {
let mut wi = wi.into_iter().collect::<Vec<_>>();
wi.sort_by_key(|(tid, info)| {
(
match info.state {
WorkerState::Busy | WorkerState::Throttled(_) => 0,
WorkerState::Idle => 1,
WorkerState::Done => 2,
},
*tid,
)
});
let mut table = vec![];
for (tid, info) in wi.iter() {
if wlo.busy && !matches!(info.state, WorkerState::Busy | WorkerState::Throttled(_)) {
continue;
}
if wlo.errors && info.errors == 0 {
continue;
}
table.push(format!("{}\t{}\t{}", tid, info.state, info.name));
if let Some(i) = &info.info {
table.push(format!("\t\t {}", i));
}
let tf = timeago::Formatter::new();
let (err_ago, err_msg) = info
.last_error
.as_ref()
.map(|(m, t)| {
(
tf.convert(Duration::from_millis(now_msec() - t)),
m.as_str(),
)
})
.unwrap_or(("(?) ago".into(), "(?)"));
if info.consecutive_errors > 0 {
table.push(format!(
"\t\t {} consecutive errors ({} total), last {}",
info.consecutive_errors, info.errors, err_ago,
));
table.push(format!("\t\t {}", err_msg));
} else if info.errors > 0 {
table.push(format!("\t\t ({} errors, last {})", info.errors, err_ago,));
if wlo.errors {
table.push(format!("\t\t {}", err_msg));
}
}
}
format_table(table);
}
+66 -12
View File
@@ -8,6 +8,14 @@ mod admin;
mod cli;
mod repair;
mod server;
#[cfg(feature = "telemetry-otlp")]
mod tracing_setup;
#[cfg(not(any(feature = "bundled-libs", feature = "system-libs")))]
compile_error!("Either bundled-libs or system-libs Cargo feature must be enabled");
#[cfg(all(feature = "bundled-libs", feature = "system-libs"))]
compile_error!("Only one of bundled-libs and system-libs Cargo features must be enabled");
use std::net::SocketAddr;
use std::path::PathBuf;
@@ -28,7 +36,10 @@ use admin::*;
use cli::*;
#[derive(StructOpt, Debug)]
#[structopt(name = "garage")]
#[structopt(
name = "garage",
about = "S3-compatible object store for self-hosted geo-distributed deployments"
)]
struct Opt {
/// Host to connect to for admin operations, in the format:
/// <public-key>@<ip>:<port>
@@ -57,20 +68,56 @@ async fn main() {
if std::env::var("RUST_LOG").is_err() {
std::env::set_var("RUST_LOG", "netapp=info,garage=info")
}
pretty_env_logger::init();
tracing_subscriber::fmt()
.with_writer(std::io::stderr)
.with_env_filter(tracing_subscriber::filter::EnvFilter::from_default_env())
.init();
sodiumoxide::init().expect("Unable to init sodiumoxide");
let opt = Opt::from_args();
// Abort on panic (same behavior as in Go)
std::panic::set_hook(Box::new(|panic_info| {
error!("{}", panic_info.to_string());
std::process::abort();
}));
// Initialize version and features info
let features = &[
#[cfg(feature = "k2v")]
"k2v",
#[cfg(feature = "sled")]
"sled",
#[cfg(feature = "lmdb")]
"lmdb",
#[cfg(feature = "sqlite")]
"sqlite",
#[cfg(feature = "kubernetes-discovery")]
"kubernetes-discovery",
#[cfg(feature = "metrics")]
"metrics",
#[cfg(feature = "telemetry-otlp")]
"telemetry-otlp",
#[cfg(feature = "bundled-libs")]
"bundled-libs",
#[cfg(feature = "system-libs")]
"system-libs",
][..];
if let Some(git_version) = option_env!("GIT_VERSION") {
garage_util::version::init_version(git_version);
}
garage_util::version::init_features(features);
// Parse arguments
let version = format!(
"{} [features: {}]",
garage_util::version::garage_version(),
features.join(", ")
);
let opt = Opt::from_clap(&Opt::clap().version(version.as_str()).get_matches());
let res = match opt.cmd {
Command::Server => {
// Abort on panic (same behavior as in Go)
std::panic::set_hook(Box::new(|panic_info| {
error!("{}", panic_info.to_string());
std::process::abort();
}));
server::run_server(opt.config_file).await
Command::Server => server::run_server(opt.config_file).await,
Command::OfflineRepair(repair_opt) => {
repair::offline::offline_repair(opt.config_file, repair_opt).await
}
Command::Node(NodeOperation::NodeId(node_id_opt)) => {
node_id_command(opt.config_file, node_id_opt.quiet)
@@ -115,7 +162,13 @@ async fn cli_command(opt: Opt) -> Result<(), Error> {
} else {
let node_id = garage_rpc::system::read_node_id(&config.as_ref().unwrap().metadata_dir)
.err_context(READ_KEY_ERROR)?;
if let Some(a) = config.as_ref().and_then(|c| c.rpc_public_addr) {
if let Some(a) = config.as_ref().and_then(|c| c.rpc_public_addr.as_ref()) {
use std::net::ToSocketAddrs;
let a = a
.to_socket_addrs()
.ok_or_message("unable to resolve rpc_public_addr specified in config file")?
.next()
.ok_or_message("unable to resolve rpc_public_addr specified in config file")?;
(node_id, a)
} else {
let default_addr = SocketAddr::new(
@@ -141,6 +194,7 @@ async fn cli_command(opt: Opt) -> Result<(), Error> {
match cli_command_dispatch(opt.cmd, &system_rpc_endpoint, &admin_rpc_endpoint, id).await {
Err(HelperError::Internal(i)) => Err(Error::Message(format!("Internal error: {}", i))),
Err(HelperError::BadRequest(b)) => Err(Error::Message(b)),
Err(e) => Err(Error::Message(format!("{}", e))),
Ok(x) => Ok(x),
}
}
-149
View File
@@ -1,149 +0,0 @@
use std::sync::Arc;
use tokio::sync::watch;
use garage_model::block_ref_table::*;
use garage_model::garage::Garage;
use garage_model::object_table::*;
use garage_model::version_table::*;
use garage_table::*;
use garage_util::error::Error;
use crate::*;
pub struct Repair {
pub garage: Arc<Garage>,
}
impl Repair {
pub async fn repair_worker(&self, opt: RepairOpt, must_exit: watch::Receiver<bool>) {
if let Err(e) = self.repair_worker_aux(opt, must_exit).await {
warn!("Repair worker failed with error: {}", e);
}
}
async fn repair_worker_aux(
&self,
opt: RepairOpt,
must_exit: watch::Receiver<bool>,
) -> Result<(), Error> {
match opt.what {
RepairWhat::Tables => {
info!("Launching a full sync of tables");
self.garage.bucket_table.syncer.add_full_sync();
self.garage.object_table.syncer.add_full_sync();
self.garage.version_table.syncer.add_full_sync();
self.garage.block_ref_table.syncer.add_full_sync();
self.garage.key_table.syncer.add_full_sync();
}
RepairWhat::Versions => {
info!("Repairing the versions table");
self.repair_versions(&must_exit).await?;
}
RepairWhat::BlockRefs => {
info!("Repairing the block refs table");
self.repair_block_ref(&must_exit).await?;
}
RepairWhat::Blocks => {
info!("Repairing the stored blocks");
self.garage
.block_manager
.repair_data_store(&must_exit)
.await?;
}
RepairWhat::Scrub { tranquility } => {
info!("Verifying integrity of stored blocks");
self.garage
.block_manager
.scrub_data_store(&must_exit, tranquility)
.await?;
}
}
Ok(())
}
async fn repair_versions(&self, must_exit: &watch::Receiver<bool>) -> Result<(), Error> {
let mut pos = vec![];
while let Some((item_key, item_bytes)) =
self.garage.version_table.data.store.get_gt(&pos)?
{
pos = item_key.to_vec();
let version = rmp_serde::decode::from_read_ref::<_, Version>(item_bytes.as_ref())?;
if version.deleted.get() {
continue;
}
let object = self
.garage
.object_table
.get(&version.bucket_id, &version.key)
.await?;
let version_exists = match object {
Some(o) => o
.versions()
.iter()
.any(|x| x.uuid == version.uuid && x.state != ObjectVersionState::Aborted),
None => false,
};
if !version_exists {
info!("Repair versions: marking version as deleted: {:?}", version);
self.garage
.version_table
.insert(&Version::new(
version.uuid,
version.bucket_id,
version.key,
true,
))
.await?;
}
if *must_exit.borrow() {
break;
}
}
Ok(())
}
async fn repair_block_ref(&self, must_exit: &watch::Receiver<bool>) -> Result<(), Error> {
let mut pos = vec![];
while let Some((item_key, item_bytes)) =
self.garage.block_ref_table.data.store.get_gt(&pos)?
{
pos = item_key.to_vec();
let block_ref = rmp_serde::decode::from_read_ref::<_, BlockRef>(item_bytes.as_ref())?;
if block_ref.deleted.get() {
continue;
}
let version = self
.garage
.version_table
.get(&block_ref.version, &EmptyKey)
.await?;
// The version might not exist if it has been GC'ed
let ref_exists = version.map(|v| !v.deleted.get()).unwrap_or(false);
if !ref_exists {
info!(
"Repair block ref: marking block_ref as deleted: {:?}",
block_ref
);
self.garage
.block_ref_table
.insert(&BlockRef {
block: block_ref.block,
version: block_ref.version,
deleted: true.into(),
})
.await?;
}
if *must_exit.borrow() {
break;
}
}
Ok(())
}
}
+2
View File
@@ -0,0 +1,2 @@
pub mod offline;
pub mod online;
+55
View File
@@ -0,0 +1,55 @@
use std::path::PathBuf;
use tokio::sync::watch;
use garage_util::background::*;
use garage_util::config::*;
use garage_util::error::*;
use garage_model::garage::Garage;
use crate::cli::structs::*;
pub async fn offline_repair(config_file: PathBuf, opt: OfflineRepairOpt) -> Result<(), Error> {
if !opt.yes {
return Err(Error::Message(
"Please add the --yes flag to launch repair operation".into(),
));
}
info!("Loading configuration...");
let config = read_config(config_file)?;
info!("Initializing background runner...");
let (done_tx, done_rx) = watch::channel(false);
let (background, await_background_done) = BackgroundRunner::new(16, done_rx);
info!("Initializing Garage main data store...");
let garage = Garage::new(config.clone(), background)?;
info!("Launching repair operation...");
match opt.what {
#[cfg(feature = "k2v")]
OfflineRepairWhat::K2VItemCounters => {
garage
.k2v
.counter_table
.offline_recount_all(&garage.k2v.item_table)?;
}
OfflineRepairWhat::ObjectCounters => {
garage
.object_counter_table
.offline_recount_all(&garage.object_table)?;
}
}
info!("Repair operation finished, shutting down Garage internals...");
done_tx.send(true).unwrap();
drop(garage);
await_background_done.await?;
info!("Cleaning up...");
Ok(())
}
+215
View File
@@ -0,0 +1,215 @@
use std::sync::Arc;
use std::time::Duration;
use async_trait::async_trait;
use tokio::sync::watch;
use garage_block::repair::ScrubWorkerCommand;
use garage_model::garage::Garage;
use garage_model::s3::block_ref_table::*;
use garage_model::s3::object_table::*;
use garage_model::s3::version_table::*;
use garage_table::*;
use garage_util::background::*;
use garage_util::error::Error;
use crate::*;
pub async fn launch_online_repair(garage: Arc<Garage>, opt: RepairOpt) {
match opt.what {
RepairWhat::Tables => {
info!("Launching a full sync of tables");
garage.bucket_table.syncer.add_full_sync();
garage.object_table.syncer.add_full_sync();
garage.version_table.syncer.add_full_sync();
garage.block_ref_table.syncer.add_full_sync();
garage.key_table.syncer.add_full_sync();
}
RepairWhat::Versions => {
info!("Repairing the versions table");
garage
.background
.spawn_worker(RepairVersionsWorker::new(garage.clone()));
}
RepairWhat::BlockRefs => {
info!("Repairing the block refs table");
garage
.background
.spawn_worker(RepairBlockrefsWorker::new(garage.clone()));
}
RepairWhat::Blocks => {
info!("Repairing the stored blocks");
garage
.background
.spawn_worker(garage_block::repair::RepairWorker::new(
garage.block_manager.clone(),
));
}
RepairWhat::Scrub { cmd } => {
let cmd = match cmd {
ScrubCmd::Start => ScrubWorkerCommand::Start,
ScrubCmd::Pause => ScrubWorkerCommand::Pause(Duration::from_secs(3600 * 24)),
ScrubCmd::Resume => ScrubWorkerCommand::Resume,
ScrubCmd::Cancel => ScrubWorkerCommand::Cancel,
ScrubCmd::SetTranquility { tranquility } => {
ScrubWorkerCommand::SetTranquility(tranquility)
}
};
info!("Sending command to scrub worker: {:?}", cmd);
garage.block_manager.send_scrub_command(cmd).await;
}
}
}
// ----
struct RepairVersionsWorker {
garage: Arc<Garage>,
pos: Vec<u8>,
counter: usize,
}
impl RepairVersionsWorker {
fn new(garage: Arc<Garage>) -> Self {
Self {
garage,
pos: vec![],
counter: 0,
}
}
}
#[async_trait]
impl Worker for RepairVersionsWorker {
fn name(&self) -> String {
"Version repair worker".into()
}
fn info(&self) -> Option<String> {
Some(format!("{} items done", self.counter))
}
async fn work(&mut self, _must_exit: &mut watch::Receiver<bool>) -> Result<WorkerState, Error> {
let item_bytes = match self.garage.version_table.data.store.get_gt(&self.pos)? {
Some((k, v)) => {
self.pos = k;
v
}
None => {
info!("repair_versions: finished, done {}", self.counter);
return Ok(WorkerState::Done);
}
};
self.counter += 1;
let version = rmp_serde::decode::from_read_ref::<_, Version>(&item_bytes)?;
if !version.deleted.get() {
let object = self
.garage
.object_table
.get(&version.bucket_id, &version.key)
.await?;
let version_exists = match object {
Some(o) => o
.versions()
.iter()
.any(|x| x.uuid == version.uuid && x.state != ObjectVersionState::Aborted),
None => false,
};
if !version_exists {
info!("Repair versions: marking version as deleted: {:?}", version);
self.garage
.version_table
.insert(&Version::new(
version.uuid,
version.bucket_id,
version.key,
true,
))
.await?;
}
}
Ok(WorkerState::Busy)
}
async fn wait_for_work(&mut self, _must_exit: &watch::Receiver<bool>) -> WorkerState {
unreachable!()
}
}
// ----
struct RepairBlockrefsWorker {
garage: Arc<Garage>,
pos: Vec<u8>,
counter: usize,
}
impl RepairBlockrefsWorker {
fn new(garage: Arc<Garage>) -> Self {
Self {
garage,
pos: vec![],
counter: 0,
}
}
}
#[async_trait]
impl Worker for RepairBlockrefsWorker {
fn name(&self) -> String {
"Block refs repair worker".into()
}
fn info(&self) -> Option<String> {
Some(format!("{} items done", self.counter))
}
async fn work(&mut self, _must_exit: &mut watch::Receiver<bool>) -> Result<WorkerState, Error> {
let item_bytes = match self.garage.block_ref_table.data.store.get_gt(&self.pos)? {
Some((k, v)) => {
self.pos = k;
v
}
None => {
info!("repair_block_ref: finished, done {}", self.counter);
return Ok(WorkerState::Done);
}
};
self.counter += 1;
let block_ref = rmp_serde::decode::from_read_ref::<_, BlockRef>(&item_bytes)?;
if !block_ref.deleted.get() {
let version = self
.garage
.version_table
.get(&block_ref.version, &EmptyKey)
.await?;
// The version might not exist if it has been GC'ed
let ref_exists = version.map(|v| !v.deleted.get()).unwrap_or(false);
if !ref_exists {
info!(
"Repair block ref: marking block_ref as deleted: {:?}",
block_ref
);
self.garage
.block_ref_table
.insert(&BlockRef {
block: block_ref.block,
version: block_ref.version,
deleted: true.into(),
})
.await?;
}
}
Ok(WorkerState::Busy)
}
async fn wait_for_work(&mut self, _must_exit: &watch::Receiver<bool>) -> WorkerState {
unreachable!()
}
}
+138 -48
View File
@@ -6,13 +6,17 @@ use garage_util::background::*;
use garage_util::config::*;
use garage_util::error::Error;
use garage_admin::metrics::*;
use garage_admin::tracing_setup::*;
use garage_api::run_api_server;
use garage_api::admin::api_server::AdminApiServer;
use garage_api::s3::api_server::S3ApiServer;
use garage_model::garage::Garage;
use garage_web::run_web_server;
use garage_web::WebServer;
#[cfg(feature = "k2v")]
use garage_api::k2v::api_server::K2VApiServer;
use crate::admin::*;
#[cfg(feature = "telemetry-otlp")]
use crate::tracing_setup::*;
async fn wait_from(mut chan: watch::Receiver<bool>) {
while !*chan.borrow() {
@@ -24,79 +28,124 @@ async fn wait_from(mut chan: watch::Receiver<bool>) {
pub async fn run_server(config_file: PathBuf) -> Result<(), Error> {
info!("Loading configuration...");
let config = read_config(config_file).expect("Unable to read config file");
let config = read_config(config_file)?;
info!("Opening database...");
let mut db_path = config.metadata_dir.clone();
db_path.push("db");
let db = sled::Config::default()
.path(&db_path)
.cache_capacity(config.sled_cache_capacity)
.flush_every_ms(Some(config.sled_flush_every_ms))
.open()
.expect("Unable to open sled DB");
// ---- Initialize Garage internals ----
info!("Initialize admin web server and metric backend...");
let admin_server_init = AdminServer::init();
#[cfg(feature = "metrics")]
let metrics_exporter = opentelemetry_prometheus::exporter().init();
info!("Initializing background runner...");
let watch_cancel = netapp::util::watch_ctrl_c();
let watch_cancel = watch_shutdown_signal();
let (background, await_background_done) = BackgroundRunner::new(16, watch_cancel.clone());
info!("Initializing Garage main data store...");
let garage = Garage::new(config.clone(), db, background);
let garage = Garage::new(config.clone(), background)?;
info!("Initialize tracing...");
if let Some(export_to) = config.admin.trace_sink {
init_tracing(&export_to, garage.system.id)?;
if config.admin.trace_sink.is_some() {
info!("Initialize tracing...");
#[cfg(feature = "telemetry-otlp")]
init_tracing(config.admin.trace_sink.as_ref().unwrap(), garage.system.id)?;
#[cfg(not(feature = "telemetry-otlp"))]
error!("Garage was built without OTLP exporter, admin.trace_sink is ignored.");
}
info!("Initialize Admin API server and metrics collector...");
let admin_server = AdminApiServer::new(
garage.clone(),
#[cfg(feature = "metrics")]
metrics_exporter,
);
info!("Launching internal Garage cluster communications...");
let run_system = tokio::spawn(garage.system.clone().run(watch_cancel.clone()));
info!("Create admin RPC handler...");
AdminRpcHandler::new(garage.clone());
info!("Initializing API server...");
let api_server = tokio::spawn(run_api_server(
garage.clone(),
wait_from(watch_cancel.clone()),
));
// ---- Launch public-facing API servers ----
info!("Initializing web server...");
let web_server = tokio::spawn(run_web_server(
garage.clone(),
wait_from(watch_cancel.clone()),
));
let mut servers = vec![];
let admin_server = if let Some(admin_bind_addr) = config.admin.api_bind_addr {
info!("Configure and run admin web server...");
Some(tokio::spawn(
admin_server_init.run(admin_bind_addr, wait_from(watch_cancel.clone())),
))
} else {
None
};
if let Some(s3_bind_addr) = &config.s3_api.api_bind_addr {
info!("Initializing S3 API server...");
servers.push((
"S3 API",
tokio::spawn(S3ApiServer::run(
garage.clone(),
*s3_bind_addr,
config.s3_api.s3_region.clone(),
wait_from(watch_cancel.clone()),
)),
));
}
if config.k2v_api.is_some() {
#[cfg(feature = "k2v")]
{
info!("Initializing K2V API server...");
servers.push((
"K2V API",
tokio::spawn(K2VApiServer::run(
garage.clone(),
config.k2v_api.as_ref().unwrap().api_bind_addr,
config.s3_api.s3_region.clone(),
wait_from(watch_cancel.clone()),
)),
));
}
#[cfg(not(feature = "k2v"))]
error!("K2V is not enabled in this build, cannot start K2V API server");
}
if let Some(web_config) = &config.s3_web {
info!("Initializing web server...");
servers.push((
"Web",
tokio::spawn(WebServer::run(
garage.clone(),
web_config.bind_addr,
web_config.root_domain.clone(),
wait_from(watch_cancel.clone()),
)),
));
}
if let Some(admin_bind_addr) = &config.admin.api_bind_addr {
info!("Launching Admin API server...");
servers.push((
"Admin",
tokio::spawn(admin_server.run(*admin_bind_addr, wait_from(watch_cancel.clone()))),
));
}
#[cfg(not(feature = "metrics"))]
if config.admin.metrics_token.is_some() {
warn!("This Garage version is built without the metrics feature");
}
// Stuff runs
// When a cancel signal is sent, stuff stops
if let Err(e) = api_server.await? {
warn!("API server exited with error: {}", e);
}
if let Err(e) = web_server.await? {
warn!("Web server exited with error: {}", e);
}
if let Some(a) = admin_server {
if let Err(e) = a.await? {
warn!("Admin web server exited with error: {}", e);
// Collect stuff
for (desc, join_handle) in servers {
if let Err(e) = join_handle.await? {
error!("{} server exited with error: {}", desc, e);
} else {
info!("{} server exited without error.", desc);
}
}
// Remove RPC handlers for system to break reference cycles
garage.system.netapp.drop_all_handlers();
opentelemetry::global::shutdown_tracer_provider();
// Await for netapp RPC system to end
run_system.await?;
info!("Netapp exited");
// Drop all references so that stuff can terminate properly
drop(garage);
@@ -108,3 +157,44 @@ pub async fn run_server(config_file: PathBuf) -> Result<(), Error> {
Ok(())
}
#[cfg(unix)]
fn watch_shutdown_signal() -> watch::Receiver<bool> {
use tokio::signal::unix::*;
let (send_cancel, watch_cancel) = watch::channel(false);
tokio::spawn(async move {
let mut sigint = signal(SignalKind::interrupt()).expect("Failed to install SIGINT handler");
let mut sigterm =
signal(SignalKind::terminate()).expect("Failed to install SIGTERM handler");
let mut sighup = signal(SignalKind::hangup()).expect("Failed to install SIGHUP handler");
tokio::select! {
_ = sigint.recv() => info!("Received SIGINT, shutting down."),
_ = sigterm.recv() => info!("Received SIGTERM, shutting down."),
_ = sighup.recv() => info!("Received SIGHUP, shutting down."),
}
send_cancel.send(true).unwrap();
});
watch_cancel
}
#[cfg(windows)]
fn watch_shutdown_signal() -> watch::Receiver<bool> {
use tokio::signal::windows::*;
let (send_cancel, watch_cancel) = watch::channel(false);
tokio::spawn(async move {
let mut sigint = ctrl_c().expect("Failed to install Ctrl-C handler");
let mut sigclose = ctrl_close().expect("Failed to install Ctrl-Close handler");
let mut siglogoff = ctrl_logoff().expect("Failed to install Ctrl-Logoff handler");
let mut sigsdown = ctrl_shutdown().expect("Failed to install Ctrl-Shutdown handler");
tokio::select! {
_ = sigint.recv() => info!("Received Ctrl-C, shutting down."),
_ = sigclose.recv() => info!("Received Ctrl-Close, shutting down."),
_ = siglogoff.recv() => info!("Received Ctrl-Logoff, shutting down."),
_ = sigsdown.recv() => info!("Received Ctrl-Shutdown, shutting down."),
}
send_cancel.send(true).unwrap();
});
watch_cancel
}
+3 -5
View File
@@ -29,8 +29,7 @@ async fn test_bucket_all() {
.unwrap()
.iter()
.filter(|x| x.name.as_ref().is_some())
.find(|x| x.name.as_ref().unwrap() == "hello")
.is_some());
.any(|x| x.name.as_ref().unwrap() == "hello"));
}
{
// Get its location
@@ -75,13 +74,12 @@ async fn test_bucket_all() {
{
// Check bucket is deleted with List buckets
let r = ctx.client.list_buckets().send().await.unwrap();
assert!(r
assert!(!r
.buckets
.as_ref()
.unwrap()
.iter()
.filter(|x| x.name.as_ref().is_some())
.find(|x| x.name.as_ref().unwrap() == "hello")
.is_none());
.any(|x| x.name.as_ref().unwrap() == "hello"));
}
}
+1 -1
View File
@@ -10,7 +10,7 @@ pub fn build_client(instance: &Instance) -> Client {
None,
"garage-integ-test",
);
let endpoint = Endpoint::immutable(instance.uri());
let endpoint = Endpoint::immutable(instance.s3_uri());
let config = Config::builder()
.region(super::REGION)
+47 -8
View File
@@ -17,14 +17,25 @@ use garage_api::signature;
pub struct CustomRequester {
key: Key,
uri: Uri,
service: &'static str,
client: Client<HttpConnector>,
}
impl CustomRequester {
pub fn new(instance: &Instance) -> Self {
pub fn new_s3(instance: &Instance) -> Self {
CustomRequester {
key: instance.key.clone(),
uri: instance.uri(),
uri: instance.s3_uri(),
service: "s3",
client: Client::new(),
}
}
pub fn new_k2v(instance: &Instance) -> Self {
CustomRequester {
key: instance.key.clone(),
uri: instance.k2v_uri(),
service: "k2v",
client: Client::new(),
}
}
@@ -32,6 +43,7 @@ impl CustomRequester {
pub fn builder(&self, bucket: String) -> RequestBuilder<'_> {
RequestBuilder {
requester: self,
service: self.service,
bucket,
method: Method::GET,
path: String::new(),
@@ -47,6 +59,7 @@ impl CustomRequester {
pub struct RequestBuilder<'a> {
requester: &'a CustomRequester,
service: &'static str,
bucket: String,
method: Method,
path: String,
@@ -59,13 +72,17 @@ pub struct RequestBuilder<'a> {
}
impl<'a> RequestBuilder<'a> {
pub fn service(&mut self, service: &'static str) -> &mut Self {
self.service = service;
self
}
pub fn method(&mut self, method: Method) -> &mut Self {
self.method = method;
self
}
pub fn path(&mut self, path: String) -> &mut Self {
self.path = path;
pub fn path(&mut self, path: impl ToString) -> &mut Self {
self.path = path.to_string();
self
}
@@ -74,16 +91,38 @@ impl<'a> RequestBuilder<'a> {
self
}
pub fn query_param<T, U>(&mut self, param: T, value: Option<U>) -> &mut Self
where
T: ToString,
U: ToString,
{
self.query_params
.insert(param.to_string(), value.as_ref().map(ToString::to_string));
self
}
pub fn signed_headers(&mut self, signed_headers: HashMap<String, String>) -> &mut Self {
self.signed_headers = signed_headers;
self
}
pub fn signed_header(&mut self, name: impl ToString, value: impl ToString) -> &mut Self {
self.signed_headers
.insert(name.to_string(), value.to_string());
self
}
pub fn unsigned_headers(&mut self, unsigned_headers: HashMap<String, String>) -> &mut Self {
self.unsigned_headers = unsigned_headers;
self
}
pub fn unsigned_header(&mut self, name: impl ToString, value: impl ToString) -> &mut Self {
self.unsigned_headers
.insert(name.to_string(), value.to_string());
self
}
pub fn body(&mut self, body: Vec<u8>) -> &mut Self {
self.body = body;
self
@@ -106,24 +145,24 @@ impl<'a> RequestBuilder<'a> {
let query = query_param_to_string(&self.query_params);
let (host, path) = if self.vhost_style {
(
format!("{}.s3.garage", self.bucket),
format!("{}.{}.garage", self.bucket, self.service),
format!("{}{}", self.path, query),
)
} else {
(
"s3.garage".to_owned(),
format!("{}.garage", self.service),
format!("{}/{}{}", self.bucket, self.path, query),
)
};
let uri = format!("{}{}", self.requester.uri, path);
let now = Utc::now();
let scope = signature::compute_scope(&now, super::REGION.as_ref());
let scope = signature::compute_scope(&now, super::REGION.as_ref(), self.service);
let mut signer = signature::signing_hmac(
&now,
&self.requester.key.secret,
super::REGION.as_ref(),
"s3",
self.service,
)
.unwrap();
let streaming_signer = signer.clone();
+24 -10
View File
@@ -22,7 +22,9 @@ pub struct Instance {
process: process::Child,
pub path: PathBuf,
pub key: Key,
pub api_port: u16,
pub s3_port: u16,
pub k2v_port: u16,
pub web_port: u16,
}
impl Instance {
@@ -58,9 +60,12 @@ rpc_secret = "{secret}"
[s3_api]
s3_region = "{region}"
api_bind_addr = "127.0.0.1:{api_port}"
api_bind_addr = "127.0.0.1:{s3_port}"
root_domain = ".s3.garage"
[k2v_api]
api_bind_addr = "127.0.0.1:{k2v_port}"
[s3_web]
bind_addr = "127.0.0.1:{web_port}"
root_domain = ".web.garage"
@@ -72,10 +77,11 @@ api_bind_addr = "127.0.0.1:{admin_port}"
path = path.display(),
secret = GARAGE_TEST_SECRET,
region = super::REGION,
api_port = port,
rpc_port = port + 1,
web_port = port + 2,
admin_port = port + 3,
s3_port = port,
k2v_port = port + 1,
rpc_port = port + 2,
web_port = port + 3,
admin_port = port + 4,
);
fs::write(path.join("config.toml"), config).expect("Could not write garage config file");
@@ -88,7 +94,7 @@ api_bind_addr = "127.0.0.1:{admin_port}"
.arg("server")
.stdout(stdout)
.stderr(stderr)
.env("RUST_LOG", "garage=info,garage_api=debug")
.env("RUST_LOG", "garage=info,garage_api=trace")
.spawn()
.expect("Could not start garage");
@@ -96,7 +102,9 @@ api_bind_addr = "127.0.0.1:{admin_port}"
process: child,
path,
key: Key::default(),
api_port: port,
s3_port: port,
k2v_port: port + 1,
web_port: port + 3,
}
}
@@ -147,8 +155,14 @@ api_bind_addr = "127.0.0.1:{admin_port}"
String::from_utf8(output.stdout).unwrap()
}
pub fn uri(&self) -> http::Uri {
format!("http://127.0.0.1:{api_port}", api_port = self.api_port)
pub fn s3_uri(&self) -> http::Uri {
format!("http://127.0.0.1:{s3_port}", s3_port = self.s3_port)
.parse()
.expect("Could not build garage endpoint URI")
}
pub fn k2v_uri(&self) -> http::Uri {
format!("http://127.0.0.1:{k2v_port}", k2v_port = self.k2v_port)
.parse()
.expect("Could not build garage endpoint URI")
}
+10 -1
View File
@@ -17,18 +17,27 @@ pub struct Context {
pub garage: &'static garage::Instance,
pub client: Client,
pub custom_request: CustomRequester,
pub k2v: K2VContext,
}
pub struct K2VContext {
pub request: CustomRequester,
}
impl Context {
fn new() -> Self {
let garage = garage::instance();
let client = client::build_client(garage);
let custom_request = CustomRequester::new(garage);
let custom_request = CustomRequester::new_s3(garage);
let k2v_request = CustomRequester::new_k2v(garage);
Context {
garage,
client,
custom_request,
k2v: K2VContext {
request: k2v_request,
},
}
}
+612
View File
@@ -0,0 +1,612 @@
use std::collections::HashMap;
use crate::common;
use assert_json_diff::assert_json_eq;
use serde_json::json;
use super::json_body;
use hyper::Method;
#[tokio::test]
async fn test_batch() {
let ctx = common::context();
let bucket = ctx.create_bucket("test-k2v-batch");
let mut values = HashMap::new();
values.insert("a", "initial test 1");
values.insert("b", "initial test 2");
values.insert("c", "initial test 3");
values.insert("d.1", "initial test 4");
values.insert("d.2", "initial test 5");
values.insert("e", "initial test 6");
let mut ct = HashMap::new();
let res = ctx
.k2v
.request
.builder(bucket.clone())
.body(
format!(
r#"[
{{"pk": "root", "sk": "a", "ct": null, "v": "{}"}},
{{"pk": "root", "sk": "b", "ct": null, "v": "{}"}},
{{"pk": "root", "sk": "c", "ct": null, "v": "{}"}},
{{"pk": "root", "sk": "d.1", "ct": null, "v": "{}"}},
{{"pk": "root", "sk": "d.2", "ct": null, "v": "{}"}},
{{"pk": "root", "sk": "e", "ct": null, "v": "{}"}}
]"#,
base64::encode(values.get(&"a").unwrap()),
base64::encode(values.get(&"b").unwrap()),
base64::encode(values.get(&"c").unwrap()),
base64::encode(values.get(&"d.1").unwrap()),
base64::encode(values.get(&"d.2").unwrap()),
base64::encode(values.get(&"e").unwrap()),
)
.into_bytes(),
)
.method(Method::POST)
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
for sk in ["a", "b", "c", "d.1", "d.2", "e"] {
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some(sk))
.signed_header("accept", "*/*")
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/octet-stream"
);
ct.insert(
sk,
res.headers()
.get("x-garage-causality-token")
.unwrap()
.to_str()
.unwrap()
.to_string(),
);
let res_body = hyper::body::to_bytes(res.into_body())
.await
.unwrap()
.to_vec();
assert_eq!(res_body, values.get(sk).unwrap().as_bytes());
}
let res = ctx
.k2v
.request
.builder(bucket.clone())
.query_param("search", Option::<&str>::None)
.body(
br#"[
{"partitionKey": "root"},
{"partitionKey": "root", "start": "c"},
{"partitionKey": "root", "start": "c", "end": "dynamite"},
{"partitionKey": "root", "start": "c", "reverse": true, "end": "a"},
{"partitionKey": "root", "start": "c", "reverse": true, "end": "azerty"},
{"partitionKey": "root", "limit": 1},
{"partitionKey": "root", "prefix": "d"}
]"#
.to_vec(),
)
.method(Method::POST)
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
let json_res = json_body(res).await;
assert_json_eq!(
json_res,
json!([
{
"partitionKey": "root",
"prefix": null,
"start": null,
"end": null,
"limit": null,
"reverse": false,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "a", "ct": ct.get("a").unwrap(), "v": [base64::encode(values.get("a").unwrap())]},
{"sk": "b", "ct": ct.get("b").unwrap(), "v": [base64::encode(values.get("b").unwrap())]},
{"sk": "c", "ct": ct.get("c").unwrap(), "v": [base64::encode(values.get("c").unwrap())]},
{"sk": "d.1", "ct": ct.get("d.1").unwrap(), "v": [base64::encode(values.get("d.1").unwrap())]},
{"sk": "d.2", "ct": ct.get("d.2").unwrap(), "v": [base64::encode(values.get("d.2").unwrap())]},
{"sk": "e", "ct": ct.get("e").unwrap(), "v": [base64::encode(values.get("e").unwrap())]}
],
"more": false,
"nextStart": null,
},
{
"partitionKey": "root",
"prefix": null,
"start": "c",
"end": null,
"limit": null,
"reverse": false,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "c", "ct": ct.get("c").unwrap(), "v": [base64::encode(values.get("c").unwrap())]},
{"sk": "d.1", "ct": ct.get("d.1").unwrap(), "v": [base64::encode(values.get("d.1").unwrap())]},
{"sk": "d.2", "ct": ct.get("d.2").unwrap(), "v": [base64::encode(values.get("d.2").unwrap())]},
{"sk": "e", "ct": ct.get("e").unwrap(), "v": [base64::encode(values.get("e").unwrap())]}
],
"more": false,
"nextStart": null,
},
{
"partitionKey": "root",
"prefix": null,
"start": "c",
"end": "dynamite",
"limit": null,
"reverse": false,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "c", "ct": ct.get("c").unwrap(), "v": [base64::encode(values.get("c").unwrap())]},
{"sk": "d.1", "ct": ct.get("d.1").unwrap(), "v": [base64::encode(values.get("d.1").unwrap())]},
{"sk": "d.2", "ct": ct.get("d.2").unwrap(), "v": [base64::encode(values.get("d.2").unwrap())]},
],
"more": false,
"nextStart": null,
},
{
"partitionKey": "root",
"prefix": null,
"start": "c",
"end": "a",
"limit": null,
"reverse": true,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "c", "ct": ct.get("c").unwrap(), "v": [base64::encode(values.get("c").unwrap())]},
{"sk": "b", "ct": ct.get("b").unwrap(), "v": [base64::encode(values.get("b").unwrap())]},
],
"more": false,
"nextStart": null,
},
{
"partitionKey": "root",
"prefix": null,
"start": "c",
"end": "azerty",
"limit": null,
"reverse": true,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "c", "ct": ct.get("c").unwrap(), "v": [base64::encode(values.get("c").unwrap())]},
{"sk": "b", "ct": ct.get("b").unwrap(), "v": [base64::encode(values.get("b").unwrap())]},
],
"more": false,
"nextStart": null,
},
{
"partitionKey": "root",
"prefix": null,
"start": null,
"end": null,
"limit": 1,
"reverse": false,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "a", "ct": ct.get("a").unwrap(), "v": [base64::encode(values.get("a").unwrap())]}
],
"more": true,
"nextStart": "b",
},
{
"partitionKey": "root",
"prefix": "d",
"start": null,
"end": null,
"limit": null,
"reverse": false,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "d.1", "ct": ct.get("d.1").unwrap(), "v": [base64::encode(values.get("d.1").unwrap())]},
{"sk": "d.2", "ct": ct.get("d.2").unwrap(), "v": [base64::encode(values.get("d.2").unwrap())]}
],
"more": false,
"nextStart": null,
},
])
);
// Insert some new values
values.insert("c'", "new test 3");
values.insert("d.1'", "new test 4");
values.insert("d.2'", "new test 5");
let res = ctx
.k2v
.request
.builder(bucket.clone())
.body(
format!(
r#"[
{{"pk": "root", "sk": "b", "ct": "{}", "v": null}},
{{"pk": "root", "sk": "c", "ct": null, "v": "{}"}},
{{"pk": "root", "sk": "d.1", "ct": "{}", "v": "{}"}},
{{"pk": "root", "sk": "d.2", "ct": null, "v": "{}"}}
]"#,
ct.get(&"b").unwrap(),
base64::encode(values.get(&"c'").unwrap()),
ct.get(&"d.1").unwrap(),
base64::encode(values.get(&"d.1'").unwrap()),
base64::encode(values.get(&"d.2'").unwrap()),
)
.into_bytes(),
)
.method(Method::POST)
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
for sk in ["b", "c", "d.1", "d.2"] {
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some(sk))
.signed_header("accept", "*/*")
.send()
.await
.unwrap();
if sk == "b" {
assert_eq!(res.status(), 204);
} else {
assert_eq!(res.status(), 200);
}
ct.insert(
sk,
res.headers()
.get("x-garage-causality-token")
.unwrap()
.to_str()
.unwrap()
.to_string(),
);
}
let res = ctx
.k2v
.request
.builder(bucket.clone())
.query_param("search", Option::<&str>::None)
.body(
br#"[
{"partitionKey": "root"},
{"partitionKey": "root", "prefix": "d"},
{"partitionKey": "root", "prefix": "d.", "end": "d.2"},
{"partitionKey": "root", "prefix": "d.", "limit": 1},
{"partitionKey": "root", "prefix": "d.", "start": "d.2", "limit": 1},
{"partitionKey": "root", "prefix": "d.", "reverse": true},
{"partitionKey": "root", "prefix": "d.", "start": "d.2", "reverse": true},
{"partitionKey": "root", "prefix": "d.", "limit": 2}
]"#
.to_vec(),
)
.method(Method::POST)
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
let json_res = json_body(res).await;
assert_json_eq!(
json_res,
json!([
{
"partitionKey": "root",
"prefix": null,
"start": null,
"end": null,
"limit": null,
"reverse": false,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "a", "ct": ct.get("a").unwrap(), "v": [base64::encode(values.get("a").unwrap())]},
{"sk": "c", "ct": ct.get("c").unwrap(), "v": [base64::encode(values.get("c").unwrap()), base64::encode(values.get("c'").unwrap())]},
{"sk": "d.1", "ct": ct.get("d.1").unwrap(), "v": [base64::encode(values.get("d.1'").unwrap())]},
{"sk": "d.2", "ct": ct.get("d.2").unwrap(), "v": [base64::encode(values.get("d.2").unwrap()), base64::encode(values.get("d.2'").unwrap())]},
{"sk": "e", "ct": ct.get("e").unwrap(), "v": [base64::encode(values.get("e").unwrap())]}
],
"more": false,
"nextStart": null,
},
{
"partitionKey": "root",
"prefix": "d",
"start": null,
"end": null,
"limit": null,
"reverse": false,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "d.1", "ct": ct.get("d.1").unwrap(), "v": [base64::encode(values.get("d.1'").unwrap())]},
{"sk": "d.2", "ct": ct.get("d.2").unwrap(), "v": [base64::encode(values.get("d.2").unwrap()), base64::encode(values.get("d.2'").unwrap())]},
],
"more": false,
"nextStart": null,
},
{
"partitionKey": "root",
"prefix": "d.",
"start": null,
"end": "d.2",
"limit": null,
"reverse": false,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "d.1", "ct": ct.get("d.1").unwrap(), "v": [base64::encode(values.get("d.1'").unwrap())]},
],
"more": false,
"nextStart": null,
},
{
"partitionKey": "root",
"prefix": "d.",
"start": null,
"end": null,
"limit": 1,
"reverse": false,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "d.1", "ct": ct.get("d.1").unwrap(), "v": [base64::encode(values.get("d.1'").unwrap())]},
],
"more": true,
"nextStart": "d.2",
},
{
"partitionKey": "root",
"prefix": "d.",
"start": "d.2",
"end": null,
"limit": 1,
"reverse": false,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "d.2", "ct": ct.get("d.2").unwrap(), "v": [base64::encode(values.get("d.2").unwrap()), base64::encode(values.get("d.2'").unwrap())]},
],
"more": false,
"nextStart": null,
},
{
"partitionKey": "root",
"prefix": "d.",
"start": null,
"end": null,
"limit": null,
"reverse": true,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "d.2", "ct": ct.get("d.2").unwrap(), "v": [base64::encode(values.get("d.2").unwrap()), base64::encode(values.get("d.2'").unwrap())]},
{"sk": "d.1", "ct": ct.get("d.1").unwrap(), "v": [base64::encode(values.get("d.1'").unwrap())]},
],
"more": false,
"nextStart": null,
},
{
"partitionKey": "root",
"prefix": "d.",
"start": "d.2",
"end": null,
"limit": null,
"reverse": true,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "d.2", "ct": ct.get("d.2").unwrap(), "v": [base64::encode(values.get("d.2").unwrap()), base64::encode(values.get("d.2'").unwrap())]},
{"sk": "d.1", "ct": ct.get("d.1").unwrap(), "v": [base64::encode(values.get("d.1'").unwrap())]},
],
"more": false,
"nextStart": null,
},
{
"partitionKey": "root",
"prefix": "d.",
"start": null,
"end": null,
"limit": 2,
"reverse": false,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "d.1", "ct": ct.get("d.1").unwrap(), "v": [base64::encode(values.get("d.1'").unwrap())]},
{"sk": "d.2", "ct": ct.get("d.2").unwrap(), "v": [base64::encode(values.get("d.2").unwrap()), base64::encode(values.get("d.2'").unwrap())]},
],
"more": false,
"nextStart": null,
},
])
);
// Test DeleteBatch
let res = ctx
.k2v
.request
.builder(bucket.clone())
.query_param("delete", Option::<&str>::None)
.body(
br#"[
{"partitionKey": "root", "start": "a", "end": "c"},
{"partitionKey": "root", "prefix": "d"}
]"#
.to_vec(),
)
.method(Method::POST)
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
let json_res = json_body(res).await;
assert_json_eq!(
json_res,
json!([
{
"partitionKey": "root",
"prefix": null,
"start": "a",
"end": "c",
"singleItem": false,
"deletedItems": 1,
},
{
"partitionKey": "root",
"prefix": "d",
"start": null,
"end": null,
"singleItem": false,
"deletedItems": 2,
},
])
);
// update our known tombstones
for sk in ["a", "b", "d.1", "d.2"] {
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some(sk))
.signed_header("accept", "application/octet-stream")
.send()
.await
.unwrap();
assert_eq!(res.status(), 204);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/octet-stream"
);
ct.insert(
sk,
res.headers()
.get("x-garage-causality-token")
.unwrap()
.to_str()
.unwrap()
.to_string(),
);
}
let res = ctx
.k2v
.request
.builder(bucket.clone())
.query_param("search", Option::<&str>::None)
.body(
br#"[
{"partitionKey": "root"},
{"partitionKey": "root", "reverse": true},
{"partitionKey": "root", "tombstones": true}
]"#
.to_vec(),
)
.method(Method::POST)
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
let json_res = json_body(res).await;
assert_json_eq!(
json_res,
json!([
{
"partitionKey": "root",
"prefix": null,
"start": null,
"end": null,
"limit": null,
"reverse": false,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "c", "ct": ct.get("c").unwrap(), "v": [base64::encode(values.get("c").unwrap()), base64::encode(values.get("c'").unwrap())]},
{"sk": "e", "ct": ct.get("e").unwrap(), "v": [base64::encode(values.get("e").unwrap())]}
],
"more": false,
"nextStart": null,
},
{
"partitionKey": "root",
"prefix": null,
"start": null,
"end": null,
"limit": null,
"reverse": true,
"conflictsOnly": false,
"tombstones": false,
"singleItem": false,
"items": [
{"sk": "e", "ct": ct.get("e").unwrap(), "v": [base64::encode(values.get("e").unwrap())]},
{"sk": "c", "ct": ct.get("c").unwrap(), "v": [base64::encode(values.get("c").unwrap()), base64::encode(values.get("c'").unwrap())]},
],
"more": false,
"nextStart": null,
},
{
"partitionKey": "root",
"prefix": null,
"start": null,
"end": null,
"limit": null,
"reverse": false,
"conflictsOnly": false,
"tombstones": true,
"singleItem": false,
"items": [
{"sk": "a", "ct": ct.get("a").unwrap(), "v": [null]},
{"sk": "b", "ct": ct.get("b").unwrap(), "v": [null]},
{"sk": "c", "ct": ct.get("c").unwrap(), "v": [base64::encode(values.get("c").unwrap()), base64::encode(values.get("c'").unwrap())]},
{"sk": "d.1", "ct": ct.get("d.1").unwrap(), "v": [null]},
{"sk": "d.2", "ct": ct.get("d.2").unwrap(), "v": [null]},
{"sk": "e", "ct": ct.get("e").unwrap(), "v": [base64::encode(values.get("e").unwrap())]},
],
"more": false,
"nextStart": null,
},
])
);
}
+141
View File
@@ -0,0 +1,141 @@
use crate::common;
use hyper::Method;
#[tokio::test]
async fn test_error_codes() {
let ctx = common::context();
let bucket = ctx.create_bucket("test-k2v-error-codes");
// Regular insert should work (code 200)
let res = ctx
.k2v
.request
.builder(bucket.clone())
.method(Method::PUT)
.path("root")
.query_param("sort_key", Some("test1"))
.body(b"Hello, world!".to_vec())
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
// Insert with trash causality token: invalid request
let res = ctx
.k2v
.request
.builder(bucket.clone())
.method(Method::PUT)
.path("root")
.query_param("sort_key", Some("test1"))
.signed_header("x-garage-causality-token", "tra$sh")
.body(b"Hello, world!".to_vec())
.send()
.await
.unwrap();
assert_eq!(res.status(), 400);
// Search without partition key: invalid request
let res = ctx
.k2v
.request
.builder(bucket.clone())
.query_param("search", Option::<&str>::None)
.body(
br#"[
{},
]"#
.to_vec(),
)
.method(Method::POST)
.send()
.await
.unwrap();
assert_eq!(res.status(), 400);
// Search with start that is not in prefix: invalid request
let res = ctx
.k2v
.request
.builder(bucket.clone())
.query_param("search", Option::<&str>::None)
.body(
br#"[
{"partition_key": "root", "prefix": "a", "start": "bx"},
]"#
.to_vec(),
)
.method(Method::POST)
.send()
.await
.unwrap();
assert_eq!(res.status(), 400);
// Search with invalid json: 400
let res = ctx
.k2v
.request
.builder(bucket.clone())
.query_param("search", Option::<&str>::None)
.body(
br#"[
{"partition_key": "root"
]"#
.to_vec(),
)
.method(Method::POST)
.send()
.await
.unwrap();
assert_eq!(res.status(), 400);
// Batch insert with invalid causality token: 400
let res = ctx
.k2v
.request
.builder(bucket.clone())
.body(
br#"[
{"pk": "root", "sk": "a", "ct": "tra$h", "v": "aGVsbG8sIHdvcmxkCg=="}
]"#
.to_vec(),
)
.method(Method::POST)
.send()
.await
.unwrap();
assert_eq!(res.status(), 400);
// Batch insert with invalid data: 400
let res = ctx
.k2v
.request
.builder(bucket.clone())
.body(
br#"[
{"pk": "root", "sk": "a", "ct": null, "v": "aGVsbG8sIHdvcmx$Cg=="}
]"#
.to_vec(),
)
.method(Method::POST)
.send()
.await
.unwrap();
assert_eq!(res.status(), 400);
// Poll with invalid causality token: 400
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("test1"))
.query_param("causality_token", Some("tra$h"))
.query_param("timeout", Some("10"))
.signed_header("accept", "application/octet-stream")
.send()
.await
.unwrap();
assert_eq!(res.status(), 400);
}
+725
View File
@@ -0,0 +1,725 @@
use std::time::Duration;
use crate::common;
use assert_json_diff::assert_json_eq;
use serde_json::json;
use super::json_body;
use hyper::Method;
#[tokio::test]
async fn test_items_and_indices() {
let ctx = common::context();
let bucket = ctx.create_bucket("test-k2v-item-and-index");
// ReadIndex -- there should be nothing
let res = ctx
.k2v
.request
.builder(bucket.clone())
.send()
.await
.unwrap();
let res_body = json_body(res).await;
assert_json_eq!(
res_body,
json!({
"prefix": null,
"start": null,
"end": null,
"limit": null,
"reverse": false,
"partitionKeys": [],
"more": false,
"nextStart": null
})
);
let content2_len = "_: hello universe".len();
let content3_len = "_: concurrent value".len();
for (i, sk) in ["a", "b", "c", "d"].iter().enumerate() {
let content = format!("{}: hello world", sk).into_bytes();
let content2 = format!("{}: hello universe", sk).into_bytes();
let content3 = format!("{}: concurrent value", sk).into_bytes();
// Put initially, no causality token
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some(sk))
.body(content.clone())
.method(Method::PUT)
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
// Get value back
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some(sk))
.signed_header("accept", "*/*")
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/octet-stream"
);
let ct = res
.headers()
.get("x-garage-causality-token")
.unwrap()
.to_str()
.unwrap()
.to_string();
let res_body = hyper::body::to_bytes(res.into_body())
.await
.unwrap()
.to_vec();
assert_eq!(res_body, content);
// ReadIndex -- now there should be some stuff
tokio::time::sleep(Duration::from_secs(1)).await;
let res = ctx
.k2v
.request
.builder(bucket.clone())
.send()
.await
.unwrap();
let res_body = json_body(res).await;
assert_json_eq!(
res_body,
json!({
"prefix": null,
"start": null,
"end": null,
"limit": null,
"reverse": false,
"partitionKeys": [
{
"pk": "root",
"entries": i+1,
"conflicts": i,
"values": i+i+1,
"bytes": i*(content2.len() + content3.len()) + content.len(),
}
],
"more": false,
"nextStart": null
})
);
// Put again, this time with causality token
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some(sk))
.signed_header("x-garage-causality-token", ct.clone())
.body(content2.clone())
.method(Method::PUT)
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
// Get value back
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some(sk))
.signed_header("accept", "*/*")
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/octet-stream"
);
let res_body = hyper::body::to_bytes(res.into_body())
.await
.unwrap()
.to_vec();
assert_eq!(res_body, content2);
// ReadIndex -- now there should be some stuff
tokio::time::sleep(Duration::from_secs(1)).await;
let res = ctx
.k2v
.request
.builder(bucket.clone())
.send()
.await
.unwrap();
let res_body = json_body(res).await;
assert_json_eq!(
res_body,
json!({
"prefix": null,
"start": null,
"end": null,
"limit": null,
"reverse": false,
"partitionKeys": [
{
"pk": "root",
"entries": i+1,
"conflicts": i,
"values": i+i+1,
"bytes": i*content3.len() + (i+1)*content2.len(),
}
],
"more": false,
"nextStart": null
})
);
// Put again with same CT, now we have concurrent values
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some(sk))
.signed_header("x-garage-causality-token", ct.clone())
.body(content3.clone())
.method(Method::PUT)
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
// Get value back
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some(sk))
.signed_header("accept", "*/*")
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/json"
);
let res_json = json_body(res).await;
assert_json_eq!(
res_json,
[base64::encode(&content2), base64::encode(&content3)]
);
// ReadIndex -- now there should be some stuff
tokio::time::sleep(Duration::from_secs(1)).await;
let res = ctx
.k2v
.request
.builder(bucket.clone())
.send()
.await
.unwrap();
let res_body = json_body(res).await;
assert_json_eq!(
res_body,
json!({
"prefix": null,
"start": null,
"end": null,
"limit": null,
"reverse": false,
"partitionKeys": [
{
"pk": "root",
"entries": i+1,
"conflicts": i+1,
"values": 2*(i+1),
"bytes": (i+1)*(content2.len() + content3.len()),
}
],
"more": false,
"nextStart": null
})
);
}
// Now delete things
for (i, sk) in ["a", "b", "c", "d"].iter().enumerate() {
// Get value back (we just need the CT)
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some(sk))
.signed_header("accept", "*/*")
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
let ct = res
.headers()
.get("x-garage-causality-token")
.unwrap()
.to_str()
.unwrap()
.to_string();
// Delete it
let res = ctx
.k2v
.request
.builder(bucket.clone())
.method(Method::DELETE)
.path("root")
.query_param("sort_key", Some(sk))
.signed_header("x-garage-causality-token", ct)
.send()
.await
.unwrap();
assert_eq!(res.status(), 204);
// ReadIndex -- now there should be some stuff
tokio::time::sleep(Duration::from_secs(1)).await;
let res = ctx
.k2v
.request
.builder(bucket.clone())
.send()
.await
.unwrap();
let res_body = json_body(res).await;
if i < 3 {
assert_json_eq!(
res_body,
json!({
"prefix": null,
"start": null,
"end": null,
"limit": null,
"reverse": false,
"partitionKeys": [
{
"pk": "root",
"entries": 3-i,
"conflicts": 3-i,
"values": 2*(3-i),
"bytes": (3-i)*(content2_len + content3_len),
}
],
"more": false,
"nextStart": null
})
);
} else {
assert_json_eq!(
res_body,
json!({
"prefix": null,
"start": null,
"end": null,
"limit": null,
"reverse": false,
"partitionKeys": [],
"more": false,
"nextStart": null
})
);
}
}
}
#[tokio::test]
async fn test_item_return_format() {
let ctx = common::context();
let bucket = ctx.create_bucket("test-k2v-item-return-format");
let single_value = b"A single value".to_vec();
let concurrent_value = b"A concurrent value".to_vec();
// -- Test with a single value --
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.body(single_value.clone())
.method(Method::PUT)
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
// f0: either
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.signed_header("accept", "*/*")
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/octet-stream"
);
let ct = res
.headers()
.get("x-garage-causality-token")
.unwrap()
.to_str()
.unwrap()
.to_string();
let res_body = hyper::body::to_bytes(res.into_body())
.await
.unwrap()
.to_vec();
assert_eq!(res_body, single_value);
// f1: not specified
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/json"
);
let res_body = json_body(res).await;
assert_json_eq!(res_body, json!([base64::encode(&single_value)]));
// f2: binary
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.signed_header("accept", "application/octet-stream")
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/octet-stream"
);
let res_body = hyper::body::to_bytes(res.into_body())
.await
.unwrap()
.to_vec();
assert_eq!(res_body, single_value);
// f3: json
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.signed_header("accept", "application/json")
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/json"
);
let res_body = json_body(res).await;
assert_json_eq!(res_body, json!([base64::encode(&single_value)]));
// -- Test with a second, concurrent value --
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.body(concurrent_value.clone())
.method(Method::PUT)
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
// f0: either
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.signed_header("accept", "*/*")
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/json"
);
let res_body = json_body(res).await;
assert_json_eq!(
res_body,
json!([
base64::encode(&single_value),
base64::encode(&concurrent_value)
])
);
// f1: not specified
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/json"
);
let res_body = json_body(res).await;
assert_json_eq!(
res_body,
json!([
base64::encode(&single_value),
base64::encode(&concurrent_value)
])
);
// f2: binary
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.signed_header("accept", "application/octet-stream")
.send()
.await
.unwrap();
assert_eq!(res.status(), 409); // CONFLICT
// f3: json
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.signed_header("accept", "application/json")
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/json"
);
let res_body = json_body(res).await;
assert_json_eq!(
res_body,
json!([
base64::encode(&single_value),
base64::encode(&concurrent_value)
])
);
// -- Delete first value, concurrently with second insert --
// -- (we now have a concurrent value and a deletion) --
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.method(Method::DELETE)
.signed_header("x-garage-causality-token", ct)
.send()
.await
.unwrap();
assert_eq!(res.status(), 204);
// f0: either
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.signed_header("accept", "*/*")
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/json"
);
let res_body = json_body(res).await;
assert_json_eq!(res_body, json!([base64::encode(&concurrent_value), null]));
// f1: not specified
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/json"
);
let ct = res
.headers()
.get("x-garage-causality-token")
.unwrap()
.to_str()
.unwrap()
.to_string();
let res_body = json_body(res).await;
assert_json_eq!(res_body, json!([base64::encode(&concurrent_value), null]));
// f2: binary
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.signed_header("accept", "application/octet-stream")
.send()
.await
.unwrap();
assert_eq!(res.status(), 409); // CONFLICT
// f3: json
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.signed_header("accept", "application/json")
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/json"
);
let res_body = json_body(res).await;
assert_json_eq!(res_body, json!([base64::encode(&concurrent_value), null]));
// -- Delete everything --
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.method(Method::DELETE)
.signed_header("x-garage-causality-token", ct)
.send()
.await
.unwrap();
assert_eq!(res.status(), 204);
// f0: either
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.signed_header("accept", "*/*")
.send()
.await
.unwrap();
assert_eq!(res.status(), 204); // NO CONTENT
// f1: not specified
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/json"
);
let res_body = json_body(res).await;
assert_json_eq!(res_body, json!([null]));
// f2: binary
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.signed_header("accept", "application/octet-stream")
.send()
.await
.unwrap();
assert_eq!(res.status(), 204); // NO CONTENT
// f3: json
let res = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("v1"))
.signed_header("accept", "application/json")
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
assert_eq!(
res.headers().get("content-type").unwrap().to_str().unwrap(),
"application/json"
);
let res_body = json_body(res).await;
assert_json_eq!(res_body, json!([null]));
}
+18
View File
@@ -0,0 +1,18 @@
pub mod batch;
pub mod errorcodes;
pub mod item;
pub mod poll;
pub mod simple;
use hyper::{Body, Response};
pub async fn json_body(res: Response<Body>) -> serde_json::Value {
let res_body: serde_json::Value = serde_json::from_slice(
&hyper::body::to_bytes(res.into_body())
.await
.unwrap()
.to_vec()[..],
)
.unwrap();
res_body
}
+98
View File
@@ -0,0 +1,98 @@
use hyper::Method;
use std::time::Duration;
use crate::common;
#[tokio::test]
async fn test_poll() {
let ctx = common::context();
let bucket = ctx.create_bucket("test-k2v-poll");
// Write initial value
let res = ctx
.k2v
.request
.builder(bucket.clone())
.method(Method::PUT)
.path("root")
.query_param("sort_key", Some("test1"))
.body(b"Initial value".to_vec())
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
// Retrieve initial value to get its causality token
let res2 = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("test1"))
.signed_header("accept", "application/octet-stream")
.send()
.await
.unwrap();
assert_eq!(res2.status(), 200);
let ct = res2
.headers()
.get("x-garage-causality-token")
.unwrap()
.to_str()
.unwrap()
.to_string();
let res2_body = hyper::body::to_bytes(res2.into_body())
.await
.unwrap()
.to_vec();
assert_eq!(res2_body, b"Initial value");
// Start poll operation
let poll = {
let bucket = bucket.clone();
let ct = ct.clone();
tokio::spawn(async move {
let ctx = common::context();
ctx.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("test1"))
.query_param("causality_token", Some(ct))
.query_param("timeout", Some("10"))
.signed_header("accept", "application/octet-stream")
.send()
.await
})
};
// Write new value that supersedes initial one
let res = ctx
.k2v
.request
.builder(bucket.clone())
.method(Method::PUT)
.path("root")
.query_param("sort_key", Some("test1"))
.signed_header("x-garage-causality-token", ct)
.body(b"New value".to_vec())
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
// Check poll finishes with correct value
let poll_res = tokio::select! {
_ = tokio::time::sleep(Duration::from_secs(10)) => panic!("poll did not terminate in time"),
res = poll => res.unwrap().unwrap(),
};
assert_eq!(poll_res.status(), 200);
let poll_res_body = hyper::body::to_bytes(poll_res.into_body())
.await
.unwrap()
.to_vec();
assert_eq!(poll_res_body, b"New value");
}
+40
View File
@@ -0,0 +1,40 @@
use crate::common;
use hyper::Method;
#[tokio::test]
async fn test_simple() {
let ctx = common::context();
let bucket = ctx.create_bucket("test-k2v-simple");
let res = ctx
.k2v
.request
.builder(bucket.clone())
.method(Method::PUT)
.path("root")
.query_param("sort_key", Some("test1"))
.body(b"Hello, world!".to_vec())
.send()
.await
.unwrap();
assert_eq!(res.status(), 200);
let res2 = ctx
.k2v
.request
.builder(bucket.clone())
.path("root")
.query_param("sort_key", Some("test1"))
.signed_header("accept", "application/octet-stream")
.send()
.await
.unwrap();
assert_eq!(res2.status(), 200);
let res2_body = hyper::body::to_bytes(res2.into_body())
.await
.unwrap()
.to_vec();
assert_eq!(res2_body, b"Hello, world!");
}
+5 -6
View File
@@ -3,9 +3,8 @@ mod common;
mod admin;
mod bucket;
mod list;
mod multipart;
mod objects;
mod simple;
mod streaming_signature;
mod website;
mod s3;
#[cfg(feature = "k2v")]
mod k2v;
+6
View File
@@ -0,0 +1,6 @@
mod list;
mod multipart;
mod objects;
mod simple;
mod streaming_signature;
mod website;
@@ -263,4 +263,13 @@ async fn test_deleteobject() {
.unwrap();
assert!(l.contents.is_none());
// Deleting a non-existing object shouldn't be a problem
ctx.client
.delete_object()
.bucket(&bucket)
.key("l-0")
.send()
.await
.unwrap();
}
@@ -35,10 +35,7 @@ async fn test_website() {
let req = || {
Request::builder()
.method("GET")
.uri(format!(
"http://127.0.0.1:{}/",
common::garage::DEFAULT_PORT + 2
))
.uri(format!("http://127.0.0.1:{}/", ctx.garage.web_port))
.header("Host", format!("{}.web.garage", BCKT_NAME))
.body(Body::empty())
.unwrap()
@@ -170,10 +167,7 @@ async fn test_website_s3_api() {
{
let req = Request::builder()
.method("GET")
.uri(format!(
"http://127.0.0.1:{}/site/",
common::garage::DEFAULT_PORT + 2
))
.uri(format!("http://127.0.0.1:{}/site/", ctx.garage.web_port))
.header("Host", format!("{}.web.garage", BCKT_NAME))
.header("Origin", "https://example.com")
.body(Body::empty())
@@ -198,7 +192,7 @@ async fn test_website_s3_api() {
.method("GET")
.uri(format!(
"http://127.0.0.1:{}/wrong.html",
common::garage::DEFAULT_PORT + 2
ctx.garage.web_port
))
.header("Host", format!("{}.web.garage", BCKT_NAME))
.body(Body::empty())
@@ -217,10 +211,7 @@ async fn test_website_s3_api() {
{
let req = Request::builder()
.method("OPTIONS")
.uri(format!(
"http://127.0.0.1:{}/site/",
common::garage::DEFAULT_PORT + 2
))
.uri(format!("http://127.0.0.1:{}/site/", ctx.garage.web_port))
.header("Host", format!("{}.web.garage", BCKT_NAME))
.header("Origin", "https://example.com")
.header("Access-Control-Request-Method", "PUT")
@@ -244,10 +235,7 @@ async fn test_website_s3_api() {
{
let req = Request::builder()
.method("OPTIONS")
.uri(format!(
"http://127.0.0.1:{}/site/",
common::garage::DEFAULT_PORT + 2
))
.uri(format!("http://127.0.0.1:{}/site/", ctx.garage.web_port))
.header("Host", format!("{}.web.garage", BCKT_NAME))
.header("Origin", "https://example.com")
.header("Access-Control-Request-Method", "DELETE")
@@ -288,10 +276,7 @@ async fn test_website_s3_api() {
{
let req = Request::builder()
.method("OPTIONS")
.uri(format!(
"http://127.0.0.1:{}/site/",
common::garage::DEFAULT_PORT + 2
))
.uri(format!("http://127.0.0.1:{}/site/", ctx.garage.web_port))
.header("Host", format!("{}.web.garage", BCKT_NAME))
.header("Origin", "https://example.com")
.header("Access-Control-Request-Method", "PUT")
@@ -319,10 +304,7 @@ async fn test_website_s3_api() {
{
let req = Request::builder()
.method("GET")
.uri(format!(
"http://127.0.0.1:{}/site/",
common::garage::DEFAULT_PORT + 2
))
.uri(format!("http://127.0.0.1:{}/site/", ctx.garage.web_port))
.header("Host", format!("{}.web.garage", BCKT_NAME))
.body(Body::empty())
.unwrap();
+37
View File
@@ -0,0 +1,37 @@
[package]
name = "k2v-client"
version = "0.0.1"
authors = ["Trinity Pointard <trinity.pointard@gmail.com>", "Alex Auvolat <alex@adnab.me>"]
edition = "2018"
license = "AGPL-3.0"
description = "Client library for the Garage K2V protocol"
repository = "https://git.deuxfleurs.fr/Deuxfleurs/garage"
readme = "../../README.md"
[dependencies]
base64 = "0.13.0"
http = "0.2.6"
log = "0.4"
rusoto_core = "0.48.0"
rusoto_credential = "0.48.0"
rusoto_signature = "0.48.0"
serde = "1.0.137"
serde_json = "1.0.81"
thiserror = "1.0.31"
tokio = "1.17.0"
# cli deps
clap = { version = "3.1.18", optional = true, features = ["derive", "env"] }
garage_util = { version = "0.8.0", path = "../util", optional = true }
[features]
cli = ["clap", "tokio/fs", "tokio/io-std", "garage_util"]
[lib]
path = "lib.rs"
[[bin]]
name = "k2v-cli"
path = "bin/k2v-cli.rs"
required-features = ["cli"]
+25
View File
@@ -0,0 +1,25 @@
Example usage:
```sh
# all these values can be provided on the cli instead
export AWS_ACCESS_KEY_ID=GK123456
export AWS_SECRET_ACCESS_KEY=0123..789
export AWS_REGION=garage
export K2V_ENDPOINT=http://172.30.2.1:3903
export K2V_BUCKET=my-bucket
cargo run --features=cli -- read-range my-partition-key --all
cargo run --features=cli -- insert my-partition-key my-sort-key --text "my string1"
cargo run --features=cli -- insert my-partition-key my-sort-key --text "my string2"
cargo run --features=cli -- insert my-partition-key my-sort-key2 --text "my string"
cargo run --features=cli -- read-range my-partition-key --all
causality=$(cargo run --features=cli -- read my-partition-key my-sort-key2 -b | head -n1)
cargo run --features=cli -- delete my-partition-key my-sort-key2 -c $causality
causality=$(cargo run --features=cli -- read my-partition-key my-sort-key -b | head -n1)
cargo run --features=cli -- insert my-partition-key my-sort-key --text "my string3" -c $causality
cargo run --features=cli -- read-range my-partition-key --all
```
+501
View File
@@ -0,0 +1,501 @@
use std::time::Duration;
use k2v_client::*;
use garage_util::formater::format_table;
use rusoto_core::credential::AwsCredentials;
use rusoto_core::Region;
use clap::{Parser, Subcommand};
/// K2V command line interface
#[derive(Parser, Debug)]
#[clap(author, version, about, long_about = None)]
struct Args {
/// Name of the region to use
#[clap(short, long, env = "AWS_REGION", default_value = "garage")]
region: String,
/// Url of the endpoint to connect to
#[clap(short, long, env = "K2V_ENDPOINT")]
endpoint: String,
/// Access key ID
#[clap(short, long, env = "AWS_ACCESS_KEY_ID")]
key_id: String,
/// Access key ID
#[clap(short, long, env = "AWS_SECRET_ACCESS_KEY")]
secret: String,
/// Bucket name
#[clap(short, long, env = "K2V_BUCKET")]
bucket: String,
#[clap(subcommand)]
command: Command,
}
#[derive(Subcommand, Debug)]
enum Command {
/// Insert a single value
Insert {
/// Partition key to insert to
partition_key: String,
/// Sort key to insert to
sort_key: String,
/// Causality of the insertion
#[clap(short, long)]
causality: Option<String>,
/// Value to insert
#[clap(flatten)]
value: Value,
},
/// Read a single value
Read {
/// Partition key to read from
partition_key: String,
/// Sort key to read from
sort_key: String,
/// Output formating
#[clap(flatten)]
output_kind: ReadOutputKind,
},
/// Watch changes on a single value
Poll {
/// Partition key to delete from
partition_key: String,
/// Sort key to delete from
sort_key: String,
/// Causality information
#[clap(short, long)]
causality: String,
/// Timeout, in seconds
#[clap(short, long)]
timeout: Option<u64>,
/// Output formating
#[clap(flatten)]
output_kind: ReadOutputKind,
},
/// Delete a single value
Delete {
/// Partition key to delete from
partition_key: String,
/// Sort key to delete from
sort_key: String,
/// Causality information
#[clap(short, long)]
causality: String,
},
/// List partition keys
ReadIndex {
/// Output formating
#[clap(flatten)]
output_kind: BatchOutputKind,
/// Output only partition keys matching this filter
#[clap(flatten)]
filter: Filter,
},
/// Read a range of sort keys
ReadRange {
/// Partition key to read from
partition_key: String,
/// Output formating
#[clap(flatten)]
output_kind: BatchOutputKind,
/// Output only sort keys matching this filter
#[clap(flatten)]
filter: Filter,
},
/// Delete a range of sort keys
DeleteRange {
/// Partition key to delete from
partition_key: String,
/// Output formating
#[clap(flatten)]
output_kind: BatchOutputKind,
/// Delete only sort keys matching this filter
#[clap(flatten)]
filter: Filter,
},
}
/// Where to read a value from
#[derive(Parser, Debug)]
#[clap(group = clap::ArgGroup::new("value").multiple(false).required(true))]
struct Value {
/// Read value from a file. use - to read from stdin
#[clap(short, long, group = "value")]
file: Option<String>,
/// Read a base64 value from commandline
#[clap(short, long, group = "value")]
b64: Option<String>,
/// Read a raw (UTF-8) value from the commandline
#[clap(short, long, group = "value")]
text: Option<String>,
}
impl Value {
async fn to_data(&self) -> Result<Vec<u8>, Error> {
if let Some(ref text) = self.text {
Ok(text.as_bytes().to_vec())
} else if let Some(ref b64) = self.b64 {
base64::decode(b64).map_err(|_| Error::Message("invalid base64 input".into()))
} else if let Some(ref path) = self.file {
use tokio::io::AsyncReadExt;
if path == "-" {
let mut file = tokio::io::stdin();
let mut vec = Vec::new();
file.read_to_end(&mut vec).await?;
Ok(vec)
} else {
let mut file = tokio::fs::File::open(path).await?;
let mut vec = Vec::new();
file.read_to_end(&mut vec).await?;
Ok(vec)
}
} else {
unreachable!("Value must have one option set")
}
}
}
#[derive(Parser, Debug)]
#[clap(group = clap::ArgGroup::new("output-kind").multiple(false).required(false))]
struct ReadOutputKind {
/// Base64 output. Conflicts are line separated, first line is causality token
#[clap(short, long, group = "output-kind")]
b64: bool,
/// Raw output. Conflicts generate error, causality token is not returned
#[clap(short, long, group = "output-kind")]
raw: bool,
/// Human formated output
#[clap(short = 'H', long, group = "output-kind")]
human: bool,
/// JSON formated output
#[clap(short, long, group = "output-kind")]
json: bool,
}
impl ReadOutputKind {
fn display_output(&self, val: CausalValue) -> ! {
use std::io::Write;
use std::process::exit;
if self.json {
let stdout = std::io::stdout();
serde_json::to_writer_pretty(stdout, &val).unwrap();
exit(0);
}
if self.raw {
let mut val = val.value;
if val.len() != 1 {
eprintln!(
"Raw mode can only read non-concurent values, found {} values, expected 1",
val.len()
);
exit(1);
}
let val = val.pop().unwrap();
match val {
K2vValue::Value(v) => {
std::io::stdout().write_all(&v).unwrap();
exit(0);
}
K2vValue::Tombstone => {
eprintln!("Expected value, found tombstone");
exit(2);
}
}
}
let causality: String = val.causality.into();
if self.b64 {
println!("{}", causality);
for val in val.value {
match val {
K2vValue::Value(v) => {
println!("{}", base64::encode(&v))
}
K2vValue::Tombstone => {
println!();
}
}
}
exit(0);
}
// human
println!("causality: {}", causality);
println!("values:");
for val in val.value {
match val {
K2vValue::Value(v) => {
if let Ok(string) = std::str::from_utf8(&v) {
println!(" utf-8: {}", string);
} else {
println!(" base64: {}", base64::encode(&v));
}
}
K2vValue::Tombstone => {
println!(" tombstone");
}
}
}
exit(0);
}
}
#[derive(Parser, Debug)]
#[clap(group = clap::ArgGroup::new("output-kind").multiple(false).required(false))]
struct BatchOutputKind {
/// Human formated output
#[clap(short = 'H', long, group = "output-kind")]
human: bool,
/// JSON formated output
#[clap(short, long, group = "output-kind")]
json: bool,
}
/// Filter for batch operations
#[derive(Parser, Debug)]
#[clap(group = clap::ArgGroup::new("filter").multiple(true).required(true))]
struct Filter {
/// Match only keys starting with this prefix
#[clap(short, long, group = "filter")]
prefix: Option<String>,
/// Match only keys lexicographically after this key (including this key itself)
#[clap(short, long, group = "filter")]
start: Option<String>,
/// Match only keys lexicographically before this key (excluding this key)
#[clap(short, long, group = "filter")]
end: Option<String>,
/// Only match the first X keys
#[clap(short, long)]
limit: Option<u64>,
/// Return keys in reverse order
#[clap(short, long)]
reverse: bool,
/// Return only keys where conflict happened
#[clap(short, long)]
conflicts_only: bool,
/// Also include keys storing only tombstones
#[clap(short, long)]
tombstones: bool,
/// Return any key
#[clap(short, long, group = "filter")]
all: bool,
}
impl Filter {
fn k2v_filter(&self) -> k2v_client::Filter<'_> {
k2v_client::Filter {
start: self.start.as_deref(),
end: self.end.as_deref(),
prefix: self.prefix.as_deref(),
limit: self.limit,
reverse: self.reverse,
}
}
}
#[tokio::main]
async fn main() -> Result<(), Error> {
let args = Args::parse();
let region = Region::Custom {
name: args.region,
endpoint: args.endpoint,
};
let creds = AwsCredentials::new(args.key_id, args.secret, None, None);
let client = K2vClient::new(region, args.bucket, creds, None)?;
match args.command {
Command::Insert {
partition_key,
sort_key,
causality,
value,
} => {
client
.insert_item(
&partition_key,
&sort_key,
value.to_data().await?,
causality.map(Into::into),
)
.await?;
}
Command::Delete {
partition_key,
sort_key,
causality,
} => {
client
.delete_item(&partition_key, &sort_key, causality.into())
.await?;
}
Command::Read {
partition_key,
sort_key,
output_kind,
} => {
let res = client.read_item(&partition_key, &sort_key).await?;
output_kind.display_output(res);
}
Command::Poll {
partition_key,
sort_key,
causality,
timeout,
output_kind,
} => {
let timeout = timeout.map(Duration::from_secs);
let res_opt = client
.poll_item(&partition_key, &sort_key, causality.into(), timeout)
.await?;
if let Some(res) = res_opt {
output_kind.display_output(res);
} else {
println!("Delay expired and value didn't change.");
}
}
Command::ReadIndex {
output_kind,
filter,
} => {
if filter.conflicts_only || filter.tombstones {
return Err(Error::Message(
"conlicts-only and tombstones are invalid for read-index".into(),
));
}
let res = client.read_index(filter.k2v_filter()).await?;
if output_kind.json {
let values = res
.items
.into_iter()
.map(|(k, v)| {
let mut value = serde_json::to_value(v).unwrap();
value
.as_object_mut()
.unwrap()
.insert("sort_key".to_owned(), k.into());
value
})
.collect::<Vec<_>>();
let json = serde_json::json!({
"next_key": res.next_start,
"values": values,
});
let stdout = std::io::stdout();
serde_json::to_writer_pretty(stdout, &json).unwrap();
} else {
if let Some(next) = res.next_start {
println!("next key: {}", next);
}
let mut to_print = Vec::new();
to_print.push(format!("key:\tentries\tconflicts\tvalues\tbytes"));
for (k, v) in res.items {
to_print.push(format!(
"{}\t{}\t{}\t{}\t{}",
k, v.entries, v.conflicts, v.values, v.bytes
));
}
format_table(to_print);
}
}
Command::ReadRange {
partition_key,
output_kind,
filter,
} => {
let op = BatchReadOp {
partition_key: &partition_key,
filter: filter.k2v_filter(),
conflicts_only: filter.conflicts_only,
tombstones: filter.tombstones,
single_item: false,
};
let mut res = client.read_batch(&[op]).await?;
let res = res.pop().unwrap();
if output_kind.json {
let values = res
.items
.into_iter()
.map(|(k, v)| {
let mut value = serde_json::to_value(v).unwrap();
value
.as_object_mut()
.unwrap()
.insert("sort_key".to_owned(), k.into());
value
})
.collect::<Vec<_>>();
let json = serde_json::json!({
"next_key": res.next_start,
"values": values,
});
let stdout = std::io::stdout();
serde_json::to_writer_pretty(stdout, &json).unwrap();
} else {
if let Some(next) = res.next_start {
println!("next key: {}", next);
}
for (key, values) in res.items {
println!("key: {}", key);
let causality: String = values.causality.into();
println!("causality: {}", causality);
for value in values.value {
match value {
K2vValue::Value(v) => {
if let Ok(string) = std::str::from_utf8(&v) {
println!(" value(utf-8): {}", string);
} else {
println!(" value(base64): {}", base64::encode(&v));
}
}
K2vValue::Tombstone => {
println!(" tombstone");
}
}
}
}
}
}
Command::DeleteRange {
partition_key,
output_kind,
filter,
} => {
let op = BatchDeleteOp {
partition_key: &partition_key,
prefix: filter.prefix.as_deref(),
start: filter.start.as_deref(),
end: filter.end.as_deref(),
single_item: false,
};
if filter.reverse
|| filter.conflicts_only
|| filter.tombstones
|| filter.limit.is_some()
{
return Err(Error::Message(
"limit, conlicts-only, reverse and tombstones are invalid for delete-range"
.into(),
));
}
let res = client.delete_batch(&[op]).await?;
if output_kind.json {
println!("{}", res[0]);
} else {
println!("deleted {} keys", res[0]);
}
}
}
Ok(())
}
+29
View File
@@ -0,0 +1,29 @@
use std::borrow::Cow;
use thiserror::Error;
/// Errors returned by this crate
#[derive(Error, Debug)]
pub enum Error {
#[error("{0}, {1}: {2} (path = {3})")]
Remote(
http::StatusCode,
Cow<'static, str>,
Cow<'static, str>,
Cow<'static, str>,
),
#[error("received invalid response: {0}")]
InvalidResponse(Cow<'static, str>),
#[error("not found")]
NotFound,
#[error("io error: {0}")]
IoError(#[from] std::io::Error),
#[error("rusoto tls error: {0}")]
RusotoTls(#[from] rusoto_core::request::TlsError),
#[error("rusoto http error: {0}")]
RusotoHttp(#[from] rusoto_core::HttpDispatchError),
#[error("deserialization error: {0}")]
Deserialization(#[from] serde_json::Error),
#[error("{0}")]
Message(Cow<'static, str>),
}
+611
View File
@@ -0,0 +1,611 @@
use std::collections::BTreeMap;
use std::time::Duration;
use http::header::{ACCEPT, CONTENT_LENGTH, CONTENT_TYPE};
use http::status::StatusCode;
use http::HeaderMap;
use log::{debug, error};
use rusoto_core::{ByteStream, DispatchSignedRequest, HttpClient};
use rusoto_credential::AwsCredentials;
use rusoto_signature::region::Region;
use rusoto_signature::signature::SignedRequest;
use serde::de::Error as DeError;
use serde::{Deserialize, Deserializer, Serialize, Serializer};
use tokio::io::AsyncReadExt;
mod error;
pub use error::Error;
const DEFAULT_TIMEOUT: Duration = Duration::from_secs(5);
const DEFAULT_POLL_TIMEOUT: Duration = Duration::from_secs(300);
const SERVICE: &str = "k2v";
const GARAGE_CAUSALITY_TOKEN: &str = "X-Garage-Causality-Token";
/// Client used to query a K2V server.
pub struct K2vClient {
region: Region,
bucket: String,
creds: AwsCredentials,
client: HttpClient,
}
impl K2vClient {
/// Create a new K2V client.
pub fn new(
region: Region,
bucket: String,
creds: AwsCredentials,
user_agent: Option<String>,
) -> Result<Self, Error> {
let mut client = HttpClient::new()?;
if let Some(ua) = user_agent {
client.local_agent_prepend(ua);
} else {
client.local_agent_prepend(format!("k2v/{}", env!("CARGO_PKG_VERSION")));
}
Ok(K2vClient {
region,
bucket,
creds,
client,
})
}
/// Perform a ReadItem request, reading the value(s) stored for a single pk+sk.
pub async fn read_item(
&self,
partition_key: &str,
sort_key: &str,
) -> Result<CausalValue, Error> {
let mut req = SignedRequest::new(
"GET",
SERVICE,
&self.region,
&format!("/{}/{}", self.bucket, partition_key),
);
req.add_param("sort_key", sort_key);
req.add_header(ACCEPT, "application/octet-stream, application/json");
let res = self.dispatch(req, None).await?;
let causality = res
.causality_token
.ok_or_else(|| Error::InvalidResponse("missing causality token".into()))?;
if res.status == StatusCode::NO_CONTENT {
return Ok(CausalValue {
causality,
value: vec![K2vValue::Tombstone],
});
}
match res.content_type.as_deref() {
Some("application/octet-stream") => Ok(CausalValue {
causality,
value: vec![K2vValue::Value(res.body)],
}),
Some("application/json") => {
let value = serde_json::from_slice(&res.body)?;
Ok(CausalValue { causality, value })
}
Some(ct) => Err(Error::InvalidResponse(
format!("invalid content type: {}", ct).into(),
)),
None => Err(Error::InvalidResponse("missing content type".into())),
}
}
/// Perform a PollItem request, waiting for the value(s) stored for a single pk+sk to be
/// updated.
pub async fn poll_item(
&self,
partition_key: &str,
sort_key: &str,
causality: CausalityToken,
timeout: Option<Duration>,
) -> Result<Option<CausalValue>, Error> {
let timeout = timeout.unwrap_or(DEFAULT_POLL_TIMEOUT);
let mut req = SignedRequest::new(
"GET",
SERVICE,
&self.region,
&format!("/{}/{}", self.bucket, partition_key),
);
req.add_param("sort_key", sort_key);
req.add_param("causality_token", &causality.0);
req.add_param("timeout", &timeout.as_secs().to_string());
req.add_header(ACCEPT, "application/octet-stream, application/json");
let res = self.dispatch(req, Some(timeout + DEFAULT_TIMEOUT)).await?;
if res.status == StatusCode::NOT_MODIFIED {
return Ok(None);
}
let causality = res
.causality_token
.ok_or_else(|| Error::InvalidResponse("missing causality token".into()))?;
if res.status == StatusCode::NO_CONTENT {
return Ok(Some(CausalValue {
causality,
value: vec![K2vValue::Tombstone],
}));
}
match res.content_type.as_deref() {
Some("application/octet-stream") => Ok(Some(CausalValue {
causality,
value: vec![K2vValue::Value(res.body)],
})),
Some("application/json") => {
let value = serde_json::from_slice(&res.body)?;
Ok(Some(CausalValue { causality, value }))
}
Some(ct) => Err(Error::InvalidResponse(
format!("invalid content type: {}", ct).into(),
)),
None => Err(Error::InvalidResponse("missing content type".into())),
}
}
/// Perform an InsertItem request, inserting a value for a single pk+sk.
pub async fn insert_item(
&self,
partition_key: &str,
sort_key: &str,
value: Vec<u8>,
causality: Option<CausalityToken>,
) -> Result<(), Error> {
let mut req = SignedRequest::new(
"PUT",
SERVICE,
&self.region,
&format!("/{}/{}", self.bucket, partition_key),
);
req.add_param("sort_key", sort_key);
req.set_payload(Some(value));
if let Some(causality) = causality {
req.add_header(GARAGE_CAUSALITY_TOKEN, &causality.0);
}
self.dispatch(req, None).await?;
Ok(())
}
/// Perform a DeleteItem request, deleting the value(s) stored for a single pk+sk.
pub async fn delete_item(
&self,
partition_key: &str,
sort_key: &str,
causality: CausalityToken,
) -> Result<(), Error> {
let mut req = SignedRequest::new(
"DELETE",
SERVICE,
&self.region,
&format!("/{}/{}", self.bucket, partition_key),
);
req.add_param("sort_key", sort_key);
req.add_header(GARAGE_CAUSALITY_TOKEN, &causality.0);
self.dispatch(req, None).await?;
Ok(())
}
/// Perform a ReadIndex request, listing partition key which have at least one associated
/// sort key, and which matches the filter.
pub async fn read_index(
&self,
filter: Filter<'_>,
) -> Result<PaginatedRange<PartitionInfo>, Error> {
let mut req =
SignedRequest::new("GET", SERVICE, &self.region, &format!("/{}", self.bucket));
filter.insert_params(&mut req);
let res = self.dispatch(req, None).await?;
let resp: ReadIndexResponse = serde_json::from_slice(&res.body)?;
let items = resp
.partition_keys
.into_iter()
.map(|ReadIndexItem { pk, info }| (pk, info))
.collect();
Ok(PaginatedRange {
items,
next_start: resp.next_start,
})
}
/// Perform an InsertBatch request, inserting multiple values at once. Note: this operation is
/// *not* atomic: it is possible for some sub-operations to fails and others to success. In
/// that case, failure is reported.
pub async fn insert_batch(&self, operations: &[BatchInsertOp<'_>]) -> Result<(), Error> {
let mut req =
SignedRequest::new("POST", SERVICE, &self.region, &format!("/{}", self.bucket));
let payload = serde_json::to_vec(operations)?;
req.set_payload(Some(payload));
self.dispatch(req, None).await?;
Ok(())
}
/// Perform a ReadBatch request, reading multiple values or range of values at once.
pub async fn read_batch(
&self,
operations: &[BatchReadOp<'_>],
) -> Result<Vec<PaginatedRange<CausalValue>>, Error> {
let mut req =
SignedRequest::new("POST", SERVICE, &self.region, &format!("/{}", self.bucket));
req.add_param("search", "");
let payload = serde_json::to_vec(operations)?;
req.set_payload(Some(payload));
let res = self.dispatch(req, None).await?;
let resp: Vec<BatchReadResponse> = serde_json::from_slice(&res.body)?;
Ok(resp
.into_iter()
.map(|e| PaginatedRange {
items: e
.items
.into_iter()
.map(|BatchReadItem { sk, ct, v }| {
(
sk,
CausalValue {
causality: ct,
value: v,
},
)
})
.collect(),
next_start: e.next_start,
})
.collect())
}
/// Perform a DeleteBatch request, deleting mutiple values or range of values at once, without
/// providing causality information.
pub async fn delete_batch(&self, operations: &[BatchDeleteOp<'_>]) -> Result<Vec<u64>, Error> {
let mut req =
SignedRequest::new("POST", SERVICE, &self.region, &format!("/{}", self.bucket));
req.add_param("delete", "");
let payload = serde_json::to_vec(operations)?;
req.set_payload(Some(payload));
let res = self.dispatch(req, None).await?;
let resp: Vec<BatchDeleteResponse> = serde_json::from_slice(&res.body)?;
Ok(resp.into_iter().map(|r| r.deleted_items).collect())
}
async fn dispatch(
&self,
mut req: SignedRequest,
timeout: Option<Duration>,
) -> Result<Response, Error> {
req.sign(&self.creds);
let mut res = self
.client
.dispatch(req, Some(timeout.unwrap_or(DEFAULT_TIMEOUT)))
.await?;
let causality_token = res
.headers
.remove(GARAGE_CAUSALITY_TOKEN)
.map(CausalityToken);
let content_type = res.headers.remove(CONTENT_TYPE);
let body = match res.status {
StatusCode::OK => read_body(&mut res.headers, res.body).await?,
StatusCode::NO_CONTENT => Vec::new(),
StatusCode::NOT_FOUND => return Err(Error::NotFound),
StatusCode::NOT_MODIFIED => Vec::new(),
s => {
let err_body = read_body(&mut res.headers, res.body)
.await
.unwrap_or_default();
let err_body_str = std::str::from_utf8(&err_body)
.map(String::from)
.unwrap_or_else(|_| base64::encode(&err_body));
if s.is_client_error() || s.is_server_error() {
error!("Error response {}: {}", res.status, err_body_str);
let err = match serde_json::from_slice::<ErrorResponse>(&err_body) {
Ok(err) => Error::Remote(
res.status,
err.code.into(),
err.message.into(),
err.path.into(),
),
Err(_) => Error::Remote(
res.status,
"unknown".into(),
err_body_str.into(),
"?".into(),
),
};
return Err(err);
} else {
let msg = format!(
"Unexpected response code {}. Response body: {}",
res.status, err_body_str
);
error!("{}", msg);
return Err(Error::InvalidResponse(msg.into()));
}
}
};
debug!(
"Response body: {}",
std::str::from_utf8(&body)
.map(String::from)
.unwrap_or_else(|_| base64::encode(&body))
);
Ok(Response {
body,
status: res.status,
causality_token,
content_type,
})
}
}
async fn read_body(headers: &mut HeaderMap<String>, body: ByteStream) -> Result<Vec<u8>, Error> {
let body_len = headers
.get(CONTENT_LENGTH)
.and_then(|h| h.parse().ok())
.unwrap_or(0);
let mut res = Vec::with_capacity(body_len);
body.into_async_read().read_to_end(&mut res).await?;
Ok(res)
}
/// An opaque token used to convey causality between operations.
#[derive(Debug, Clone, PartialEq, Eq, Deserialize, Serialize)]
#[serde(transparent)]
pub struct CausalityToken(String);
impl From<String> for CausalityToken {
fn from(v: String) -> Self {
CausalityToken(v)
}
}
impl From<CausalityToken> for String {
fn from(v: CausalityToken) -> Self {
v.0
}
}
/// A value in K2V. can be either a binary value, or a tombstone.
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum K2vValue {
Tombstone,
Value(Vec<u8>),
}
impl From<Vec<u8>> for K2vValue {
fn from(v: Vec<u8>) -> Self {
K2vValue::Value(v)
}
}
impl From<Option<Vec<u8>>> for K2vValue {
fn from(v: Option<Vec<u8>>) -> Self {
match v {
Some(v) => K2vValue::Value(v),
None => K2vValue::Tombstone,
}
}
}
impl<'de> Deserialize<'de> for K2vValue {
fn deserialize<D>(d: D) -> Result<Self, D::Error>
where
D: Deserializer<'de>,
{
let val: Option<&str> = Option::deserialize(d)?;
Ok(match val {
Some(s) => {
K2vValue::Value(base64::decode(s).map_err(|_| DeError::custom("invalid base64"))?)
}
None => K2vValue::Tombstone,
})
}
}
impl Serialize for K2vValue {
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: Serializer,
{
match self {
K2vValue::Tombstone => serializer.serialize_none(),
K2vValue::Value(v) => {
let b64 = base64::encode(v);
serializer.serialize_str(&b64)
}
}
}
}
/// A set of K2vValue and associated causality information.
#[derive(Debug, Clone, Serialize)]
pub struct CausalValue {
pub causality: CausalityToken,
pub value: Vec<K2vValue>,
}
/// Result of paginated requests.
#[derive(Debug, Clone)]
pub struct PaginatedRange<V> {
pub items: BTreeMap<String, V>,
pub next_start: Option<String>,
}
/// Filter for batch operations.
#[derive(Debug, Default, Clone, Deserialize, Serialize)]
pub struct Filter<'a> {
pub start: Option<&'a str>,
pub end: Option<&'a str>,
pub prefix: Option<&'a str>,
pub limit: Option<u64>,
#[serde(default)]
pub reverse: bool,
}
impl<'a> Filter<'a> {
fn insert_params(&self, req: &mut SignedRequest) {
if let Some(start) = &self.start {
req.add_param("start", start);
}
if let Some(end) = &self.end {
req.add_param("end", end);
}
if let Some(prefix) = &self.prefix {
req.add_param("prefix", prefix);
}
if let Some(limit) = &self.limit {
req.add_param("limit", &limit.to_string());
}
if self.reverse {
req.add_param("reverse", "true");
}
}
}
#[derive(Debug, Clone, Deserialize)]
#[serde(rename_all = "camelCase")]
struct ReadIndexResponse<'a> {
#[serde(flatten, borrow)]
#[allow(dead_code)]
filter: Filter<'a>,
partition_keys: Vec<ReadIndexItem>,
#[allow(dead_code)]
more: bool,
next_start: Option<String>,
}
#[derive(Debug, Clone, Deserialize)]
struct ReadIndexItem {
pk: String,
#[serde(flatten)]
info: PartitionInfo,
}
/// Information about data stored with a given partition key.
#[derive(Debug, Clone, Deserialize, Serialize)]
pub struct PartitionInfo {
pub entries: u64,
pub conflicts: u64,
pub values: u64,
pub bytes: u64,
}
/// Single sub-operation of an InsertBatch.
#[derive(Debug, Clone, Serialize)]
pub struct BatchInsertOp<'a> {
#[serde(rename = "pk")]
pub partition_key: &'a str,
#[serde(rename = "sk")]
pub sort_key: &'a str,
#[serde(rename = "ct")]
pub causality: Option<CausalityToken>,
#[serde(rename = "v")]
pub value: K2vValue,
}
/// Single sub-operation of a ReadBatch.
#[derive(Debug, Default, Clone, Deserialize, Serialize)]
#[serde(rename_all = "camelCase")]
pub struct BatchReadOp<'a> {
pub partition_key: &'a str,
#[serde(flatten, borrow)]
pub filter: Filter<'a>,
#[serde(default)]
pub single_item: bool,
#[serde(default)]
pub conflicts_only: bool,
#[serde(default)]
pub tombstones: bool,
}
#[derive(Debug, Clone, Deserialize)]
#[serde(rename_all = "camelCase")]
struct BatchReadResponse<'a> {
#[serde(flatten, borrow)]
#[allow(dead_code)]
op: BatchReadOp<'a>,
items: Vec<BatchReadItem>,
#[allow(dead_code)]
more: bool,
next_start: Option<String>,
}
#[derive(Debug, Clone, Deserialize)]
struct BatchReadItem {
sk: String,
ct: CausalityToken,
v: Vec<K2vValue>,
}
/// Single sub-operation of a DeleteBatch
#[derive(Debug, Clone, Deserialize, Serialize)]
#[serde(rename_all = "camelCase")]
pub struct BatchDeleteOp<'a> {
pub partition_key: &'a str,
pub prefix: Option<&'a str>,
pub start: Option<&'a str>,
pub end: Option<&'a str>,
#[serde(default)]
pub single_item: bool,
}
impl<'a> BatchDeleteOp<'a> {
pub fn new(partition_key: &'a str) -> Self {
BatchDeleteOp {
partition_key,
prefix: None,
start: None,
end: None,
single_item: false,
}
}
}
#[derive(Debug, Clone, Deserialize)]
#[serde(rename_all = "camelCase")]
struct BatchDeleteResponse<'a> {
#[serde(flatten, borrow)]
#[allow(dead_code)]
filter: BatchDeleteOp<'a>,
deleted_items: u64,
}
#[derive(Deserialize)]
struct ErrorResponse {
code: String,
message: String,
#[allow(dead_code)]
region: String,
path: String,
}
struct Response {
body: Vec<u8>,
status: StatusCode,
causality_token: Option<CausalityToken>,
content_type: Option<String>,
}
+15 -11
View File
@@ -1,6 +1,6 @@
[package]
name = "garage_model"
version = "0.7.0"
version = "0.8.0"
authors = ["Alex Auvolat <alex@adnab.me>"]
edition = "2018"
license = "AGPL-3.0"
@@ -14,22 +14,22 @@ path = "lib.rs"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
garage_rpc = { version = "0.7.0", path = "../rpc" }
garage_table = { version = "0.7.0", path = "../table" }
garage_block = { version = "0.7.0", path = "../block" }
garage_util = { version = "0.7.0", path = "../util" }
garage_model_050 = { package = "garage_model", version = "0.5.1" }
garage_db = { version = "0.8.0", path = "../db" }
garage_rpc = { version = "0.8.0", path = "../rpc" }
garage_table = { version = "0.8.0", path = "../table" }
garage_block = { version = "0.8.0", path = "../block" }
garage_util = { version = "0.8.0", path = "../util" }
async-trait = "0.1.7"
arc-swap = "1.0"
blake2 = "0.9"
err-derive = "0.3"
hex = "0.4"
base64 = "0.13"
tracing = "0.1.30"
rand = "0.8"
zstd = { version = "0.9", default-features = false }
sled = "0.34"
rmp-serde = "0.15"
serde = { version = "1.0", default-features = false, features = ["derive", "rc"] }
serde_bytes = "0.11"
@@ -39,6 +39,10 @@ futures-util = "0.3"
tokio = { version = "1.0", default-features = false, features = ["rt", "rt-multi-thread", "io-util", "net", "time", "macros", "sync", "signal", "fs"] }
opentelemetry = "0.17"
#netapp = { version = "0.3.0", git = "https://git.deuxfleurs.fr/lx/netapp" }
#netapp = { version = "0.4", path = "../../../netapp" }
netapp = "0.4"
netapp = "0.5"
[features]
k2v = [ "garage_util/k2v" ]
lmdb = [ "garage_db/lmdb" ]
sled = [ "garage_db/sled" ]
sqlite = [ "garage_db/sqlite" ]
+1 -1
View File
@@ -7,7 +7,7 @@ use garage_table::*;
/// The bucket alias table holds the names given to buckets
/// in the global namespace.
#[derive(PartialEq, Clone, Debug, Serialize, Deserialize)]
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct BucketAlias {
name: String,
pub state: crdt::Lww<Option<Uuid>>,
+20 -3
View File
@@ -1,6 +1,6 @@
use serde::{Deserialize, Serialize};
use garage_table::crdt::Crdt;
use garage_table::crdt::*;
use garage_table::*;
use garage_util::data::*;
use garage_util::time::*;
@@ -12,7 +12,7 @@ use crate::permission::BucketKeyPerm;
/// Its parameters are not directly accessible as:
/// - It must be possible to merge paramaters, hence the use of a LWW CRDT.
/// - A bucket has 2 states, Present or Deleted and parameters make sense only if present.
#[derive(PartialEq, Clone, Debug, Serialize, Deserialize)]
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Bucket {
/// ID of the bucket
pub id: Uuid,
@@ -21,7 +21,7 @@ pub struct Bucket {
}
/// Configuration for a bucket
#[derive(PartialEq, Clone, Debug, Serialize, Deserialize)]
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct BucketParams {
/// Bucket's creation date
pub creation_date: u64,
@@ -44,6 +44,9 @@ pub struct BucketParams {
pub website_config: crdt::Lww<Option<WebsiteConfig>>,
/// CORS rules
pub cors_config: crdt::Lww<Option<Vec<CorsRule>>>,
/// Bucket quotas
#[serde(default)]
pub quotas: crdt::Lww<BucketQuotas>,
}
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
@@ -62,6 +65,18 @@ pub struct CorsRule {
pub expose_headers: Vec<String>,
}
#[derive(Default, PartialEq, Eq, PartialOrd, Ord, Clone, Debug, Serialize, Deserialize)]
pub struct BucketQuotas {
/// Maximum size in bytes (bucket size = sum of sizes of objects in the bucket)
pub max_size: Option<u64>,
/// Maximum number of non-deleted objects in the bucket
pub max_objects: Option<u64>,
}
impl AutoCrdt for BucketQuotas {
const WARN_IF_DIFFERENT: bool = true;
}
impl BucketParams {
/// Create an empty BucketParams with no authorized keys and no website accesss
pub fn new() -> Self {
@@ -72,6 +87,7 @@ impl BucketParams {
local_aliases: crdt::LwwMap::new(),
website_config: crdt::Lww::new(None),
cors_config: crdt::Lww::new(None),
quotas: crdt::Lww::new(BucketQuotas::default()),
}
}
}
@@ -86,6 +102,7 @@ impl Crdt for BucketParams {
self.website_config.merge(&o.website_config);
self.cors_config.merge(&o.cors_config);
self.quotas.merge(&o.quotas);
}
}
+175 -28
View File
@@ -2,8 +2,11 @@ use std::sync::Arc;
use netapp::NetworkKey;
use garage_db as db;
use garage_util::background::*;
use garage_util::config::*;
use garage_util::error::*;
use garage_rpc::system::System;
@@ -13,13 +16,18 @@ use garage_table::replication::TableFullReplication;
use garage_table::replication::TableShardedReplication;
use garage_table::*;
use crate::block_ref_table::*;
use crate::s3::block_ref_table::*;
use crate::s3::object_table::*;
use crate::s3::version_table::*;
use crate::bucket_alias_table::*;
use crate::bucket_table::*;
use crate::helper;
use crate::index_counter::*;
use crate::key_table::*;
use crate::object_table::*;
use crate::version_table::*;
#[cfg(feature = "k2v")]
use crate::k2v::{item_table::*, poll::*, rpc::*};
/// An entire Garage full of data
pub struct Garage {
@@ -27,7 +35,7 @@ pub struct Garage {
pub config: Config,
/// The local database
pub db: sled::Db,
pub db: db::Db,
/// A background job runner
pub background: Arc<BackgroundRunner>,
/// The membership manager
@@ -35,21 +43,118 @@ pub struct Garage {
/// The block manager
pub block_manager: Arc<BlockManager>,
/// Table containing informations about buckets
/// Table containing buckets
pub bucket_table: Arc<Table<BucketTable, TableFullReplication>>,
/// Table containing informations about bucket aliases
/// Table containing bucket aliases
pub bucket_alias_table: Arc<Table<BucketAliasTable, TableFullReplication>>,
/// Table containing informations about api keys
/// Table containing api keys
pub key_table: Arc<Table<KeyTable, TableFullReplication>>,
/// Table containing S3 objects
pub object_table: Arc<Table<ObjectTable, TableShardedReplication>>,
/// Counting table containing object counters
pub object_counter_table: Arc<IndexCounter<Object>>,
/// Table containing S3 object versions
pub version_table: Arc<Table<VersionTable, TableShardedReplication>>,
/// Table containing S3 block references (not blocks themselves)
pub block_ref_table: Arc<Table<BlockRefTable, TableShardedReplication>>,
#[cfg(feature = "k2v")]
pub k2v: GarageK2V,
}
#[cfg(feature = "k2v")]
pub struct GarageK2V {
/// Table containing K2V items
pub item_table: Arc<Table<K2VItemTable, TableShardedReplication>>,
/// Indexing table containing K2V item counters
pub counter_table: Arc<IndexCounter<K2VItem>>,
/// K2V RPC handler
pub rpc: Arc<K2VRpcHandler>,
}
impl Garage {
/// Create and run garage
pub fn new(config: Config, db: sled::Db, background: Arc<BackgroundRunner>) -> Arc<Self> {
pub fn new(config: Config, background: Arc<BackgroundRunner>) -> Result<Arc<Self>, Error> {
// Create meta dir and data dir if they don't exist already
std::fs::create_dir_all(&config.metadata_dir)
.ok_or_message("Unable to create Garage metadata directory")?;
std::fs::create_dir_all(&config.data_dir)
.ok_or_message("Unable to create Garage data directory")?;
info!("Opening database...");
let mut db_path = config.metadata_dir.clone();
let db = match config.db_engine.as_str() {
// ---- Sled DB ----
#[cfg(feature = "sled")]
"sled" => {
db_path.push("db");
info!("Opening Sled database at: {}", db_path.display());
let db = db::sled_adapter::sled::Config::default()
.path(&db_path)
.cache_capacity(config.sled_cache_capacity)
.flush_every_ms(Some(config.sled_flush_every_ms))
.open()
.expect("Unable to open sled DB");
db::sled_adapter::SledDb::init(db)
}
#[cfg(not(feature = "sled"))]
"sled" => return Err(Error::Message("sled db not available in this build".into())),
// ---- Sqlite DB ----
#[cfg(feature = "sqlite")]
"sqlite" | "sqlite3" | "rusqlite" => {
db_path.push("db.sqlite");
info!("Opening Sqlite database at: {}", db_path.display());
let db = db::sqlite_adapter::rusqlite::Connection::open(db_path)
.expect("Unable to open sqlite DB");
db::sqlite_adapter::SqliteDb::init(db)
}
#[cfg(not(feature = "sqlite"))]
"sqlite" | "sqlite3" | "rusqlite" => {
return Err(Error::Message(
"sqlite db not available in this build".into(),
))
}
// ---- LMDB DB ----
#[cfg(feature = "lmdb")]
"lmdb" | "heed" => {
db_path.push("db.lmdb");
info!("Opening LMDB database at: {}", db_path.display());
std::fs::create_dir_all(&db_path).expect("Unable to create LMDB data directory");
let map_size = garage_db::lmdb_adapter::recommended_map_size();
use db::lmdb_adapter::heed;
let mut env_builder = heed::EnvOpenOptions::new();
env_builder.max_dbs(100);
env_builder.max_readers(500);
env_builder.map_size(map_size);
unsafe {
env_builder.flag(heed::flags::Flags::MdbNoSync);
env_builder.flag(heed::flags::Flags::MdbNoMetaSync);
}
let db = env_builder.open(&db_path).expect("Unable to open LMDB DB");
db::lmdb_adapter::LmdbDb::init(db)
}
#[cfg(not(feature = "lmdb"))]
"lmdb" | "heed" => return Err(Error::Message("lmdb db not available in this build".into())),
// ---- Unavailable DB engine ----
e => {
return Err(Error::Message(format!(
"Unsupported DB engine: {} (options: {})",
e,
vec![
#[cfg(feature = "sled")]
"sled",
#[cfg(feature = "sqlite")]
"sqlite",
#[cfg(feature = "lmdb")]
"lmdb",
]
.join(", ")
)));
}
};
let network_key = NetworkKey::from_slice(
&hex::decode(&config.rpc_secret).expect("Invalid RPC secret key")[..],
)
@@ -64,7 +169,7 @@ impl Garage {
background.clone(),
replication_mode.replication_factor(),
&config,
);
)?;
let data_rep_param = TableShardedReplication {
system: system.clone(),
@@ -90,11 +195,25 @@ impl Garage {
&db,
config.data_dir.clone(),
config.compression_level,
config.block_manager_background_tranquility,
data_rep_param,
system.clone(),
);
// ---- admin tables ----
info!("Initialize bucket_table...");
let bucket_table = Table::new(BucketTable, control_rep_param.clone(), system.clone(), &db);
info!("Initialize bucket_alias_table...");
let bucket_alias_table = Table::new(
BucketAliasTable,
control_rep_param.clone(),
system.clone(),
&db,
);
info!("Initialize key_table_table...");
let key_table = Table::new(KeyTable, control_rep_param, system.clone(), &db);
// ---- S3 tables ----
info!("Initialize block_ref_table...");
let block_ref_table = Table::new(
BlockRefTable {
@@ -116,34 +235,28 @@ impl Garage {
&db,
);
info!("Initialize object counter table...");
let object_counter_table = IndexCounter::new(system.clone(), meta_rep_param.clone(), &db);
info!("Initialize object_table...");
#[allow(clippy::redundant_clone)]
let object_table = Table::new(
ObjectTable {
background: background.clone(),
version_table: version_table.clone(),
object_counter_table: object_counter_table.clone(),
},
meta_rep_param,
meta_rep_param.clone(),
system.clone(),
&db,
);
info!("Initialize bucket_table...");
let bucket_table = Table::new(BucketTable, control_rep_param.clone(), system.clone(), &db);
// ---- K2V ----
#[cfg(feature = "k2v")]
let k2v = GarageK2V::new(system.clone(), &db, meta_rep_param);
info!("Initialize bucket_alias_table...");
let bucket_alias_table = Table::new(
BucketAliasTable,
control_rep_param.clone(),
system.clone(),
&db,
);
info!("Initialize key_table_table...");
let key_table = Table::new(KeyTable, control_rep_param, system.clone(), &db);
info!("Initialize Garage...");
Arc::new(Self {
// -- done --
Ok(Arc::new(Self {
config,
db,
background,
@@ -153,12 +266,46 @@ impl Garage {
bucket_alias_table,
key_table,
object_table,
object_counter_table,
version_table,
block_ref_table,
})
#[cfg(feature = "k2v")]
k2v,
}))
}
pub fn bucket_helper(&self) -> helper::bucket::BucketHelper {
helper::bucket::BucketHelper(self)
}
pub fn key_helper(&self) -> helper::key::KeyHelper {
helper::key::KeyHelper(self)
}
}
#[cfg(feature = "k2v")]
impl GarageK2V {
fn new(system: Arc<System>, db: &db::Db, meta_rep_param: TableShardedReplication) -> Self {
info!("Initialize K2V counter table...");
let counter_table = IndexCounter::new(system.clone(), meta_rep_param.clone(), db);
info!("Initialize K2V subscription manager...");
let subscriptions = Arc::new(SubscriptionManager::new());
info!("Initialize K2V item table...");
let item_table = Table::new(
K2VItemTable {
counter_table: counter_table.clone(),
subscriptions: subscriptions.clone(),
},
meta_rep_param,
system.clone(),
db,
);
let rpc = K2VRpcHandler::new(system, item_table.clone(), subscriptions);
Self {
item_table,
counter_table,
rpc,
}
}
}

Some files were not shown because too many files have changed in this diff Show More