Compare commits

..

83 Commits

Author SHA1 Message Date
Alex Auvolat 32715d462e history table refactoring 2023-01-09 15:00:00 +01:00
Alex Auvolat b337895fce Update Rust Nix toolchain 2023-01-09 13:21:37 +01:00
Alex Auvolat 49b5d18554 K2V history and preparation for range watch 2023-01-05 13:11:48 +01:00
Alex 02e8eb167e Merge pull request 'PutObject: better cleanup when request is interrupted in the middle' (#462) from interrupted-cleanup into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/462
2023-01-04 14:43:45 +00:00
Alex 329c0e64f9 Merge pull request 'Improve garage worker set and add garage worker get' (#464) from worker-get into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/464
2023-01-04 13:47:42 +00:00
Alex Auvolat 29dbcb8278 bg var operation on all nodes at once 2023-01-04 13:25:57 +01:00
Alex Auvolat f3f27293df Uniform framework for bg variable management 2023-01-04 13:07:13 +01:00
Alex Auvolat 13c5549886 Remove token_bucket.rs 2023-01-04 11:47:56 +01:00
Alex Auvolat 936b6cb563 When saving block, delete .tmp file if we could not complete 2023-01-03 17:34:26 +01:00
Alex Auvolat 0650a43cf1 PutObject: better cleanup on Drop (incl. when request is interrupted in the middle) 2023-01-03 17:05:17 +01:00
Alex 4eb8ca3a52 Merge pull request 'Fix Consul & Kubernetes discovery with new way of doing background things' (#463) from fix-background into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/463
2023-01-03 16:04:40 +00:00
Alex Auvolat 1fc220886a Fix Consul & Kubernetes discovery with new way of doing background things 2023-01-03 16:55:59 +01:00
Alex 73ed9c7403 Merge pull request 'Refactor how things are migrated' (#461) from format-migration into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/461
2023-01-03 15:28:24 +00:00
Alex Auvolat 1d5bdc17a4 use impossible enum type 2023-01-03 16:04:06 +01:00
Alex Auvolat c106304b9c more idiomatic and shorter 2023-01-03 16:00:19 +01:00
Alex Auvolat 33f25d26c7 fix doc and add tests for migrate.rs 2023-01-03 15:53:13 +01:00
Alex Auvolat d6d571d512 cargo fmt 2023-01-03 15:30:21 +01:00
Alex Auvolat a54b67740d move debug_serialize to garage_util::encode 2023-01-03 15:29:29 +01:00
Alex Auvolat 8d5505514f Make it explicit when using nonversioned encoding 2023-01-03 15:27:36 +01:00
Alex Auvolat 426d8784da cleanup 2023-01-03 15:08:37 +01:00
Alex Auvolat a81200d345 Update cargo.nix 2023-01-03 14:45:47 +01:00
Alex Auvolat cdb2a591e9 Refactor how things are migrated 2023-01-03 14:44:47 +01:00
Alex 582b076179 Merge pull request 'Some improvements to Garage internals' (#451) from internals-rework into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/451
2023-01-03 11:37:31 +00:00
Alex Auvolat 939a6d67e8 Merge branch 'main' into internals-rework 2023-01-02 15:07:44 +01:00
Alex 76230f2028 Merge pull request 'Bump everything to v0.8.1' (#458) from up-v0.8.1 into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/458
2023-01-02 13:32:45 +00:00
Alex Auvolat 6775569525 Bump everything to v0.8.1 2023-01-02 14:15:33 +01:00
Alex Auvolat 6b857a9b8c cargo fmt 2023-01-02 13:50:42 +01:00
Alex 1649002e2b Merge pull request 'Add a note about Peertube 5.0 private videos' (#456) from kaiyou/garage:docs-apps into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/456
2023-01-02 12:49:14 +00:00
Alex 822e344845 Merge pull request 'Add some docs about using Python Minio SDK' (#455) from kaiyou/garage:docs-s3-libs into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/455
2023-01-02 12:48:52 +00:00
Alex 7f7d53cfa9 Merge pull request 'improvements to CLI and new debug features' (#448) from cli-improvements into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/448
2023-01-02 12:42:24 +00:00
kaiyou fd10200bec Add a note about Peertube 5.0 private videos 2022-12-25 14:20:01 +01:00
kaiyou 0c7ed0b0af Add some docs about using Python Minio SDK 2022-12-25 13:55:12 +01:00
Alex 1af4a5ed56 Merge pull request 'Fix router keyword handling (fix #442)' (#446) from router-keywords-fix into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/446
2022-12-15 08:40:26 +00:00
Alex Auvolat 1fcd0b371b online repair workers: retry on error 2022-12-14 16:31:31 +01:00
Alex Auvolat 13c8662126 factorize 2022-12-14 16:16:55 +01:00
Alex Auvolat e6f14ab5cf better error message handling 2022-12-14 16:11:19 +01:00
Alex Auvolat 510b620108 Get rid of background::spawn 2022-12-14 16:08:05 +01:00
Alex Auvolat dfc131850a Simplified and more aggressive worker exit logic 2022-12-14 15:25:29 +01:00
Alex Auvolat d4af27f920 Add missing notify 2022-12-14 13:54:21 +01:00
Alex Auvolat 0d6b05bb6c Update cargo.nix 2022-12-14 12:58:24 +01:00
Alex Auvolat a19bfef508 Improve error message on rpc connection failure 2022-12-14 12:57:33 +01:00
Alex Auvolat d56c472712 Refactor background runner and get rid of job worker 2022-12-14 12:51:42 +01:00
Alex Auvolat 2183518edc Spawn all background workers in a separate step 2022-12-14 12:28:07 +01:00
Alex Auvolat 83c8467e23 Proper queueing for delayed inserts, now backed to disk 2022-12-14 11:58:06 +01:00
Alex Auvolat f8e528c15d Small refactor of tables internals 2022-12-14 10:48:49 +01:00
Alex Auvolat d1279e04f3 Fix error messages 2022-12-13 16:18:01 +01:00
Alex Auvolat 041b60ed1d Add block.rc_size, table.size and table.merkle_tree_size metrics 2022-12-13 15:54:03 +01:00
Alex Auvolat f8d5409894 cli: more info displayed on error in garage stats 2022-12-13 15:46:04 +01:00
Alex Auvolat d6040e32a6 cli: prettier table in garage stats 2022-12-13 15:43:22 +01:00
Alex Auvolat d7f90cabb0 Implement block retry-now and block purge 2022-12-13 15:02:42 +01:00
Alex Auvolat 687660b27f Implement block list-errors and block info 2022-12-13 14:23:45 +01:00
Alex Auvolat 9d82196945 cli: new worker info command 2022-12-13 12:24:30 +01:00
Alex Auvolat a51e8d94c6 cli: rename resync-n-workers into resync-worker-count 2022-12-13 11:44:11 +01:00
Alex Auvolat de9d6cddf7 Prettier worker list table; remove useless CLI log messages 2022-12-12 17:17:05 +01:00
Alex f7c65e830e Merge pull request 'Properly enforce allow_create_bucket' (#447) from fix-allow-create-bucket into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/447
2022-12-12 14:55:12 +00:00
Alex Auvolat 0e61e3b6fb Fix bucket creation tests to take permissions into account 2022-12-12 15:47:55 +01:00
Alex Auvolat a0abf41762 Fix router keyword handling (fix #442) 2022-12-12 12:05:37 +01:00
Alex Auvolat 2ac75018a1 Properly enforce allow_create_bucket 2022-12-12 12:03:54 +01:00
Maximilien R 980572a887 Merge pull request 'helm: ingress improvements' (#422) from patrickjahns:helm-refactor-ingress into main
As discussed in the chat yesterday, I want to propose to disable the ingress per default.

The motivation behind this change is, that per default the ingress is "misconfigured"
meaning it can not work with the default values and requires a user of the chart to
add additional configuration. When installing the chart per default, I would not
expect to already expose garage publicly without my explicit configuration to do so

Commenting the ingressClass resource also allows for relying only on
annotations - otherwise the ingressClass would be always set to nginx
or require a user to override it with ingressClass: null

A small change on top, I've added the ability to specify user defined labels per ingress
2022-12-12 00:53:57 +01:00
Patrick Jahns 7a0014b6f7 chore(helm): bump chart number 2022-12-11 23:11:56 +00:00
Patrick Jahns edb0b9c1ee feat(helm): allow to add custom labels to created ingress resources 2022-12-11 23:11:56 +00:00
Patrick Jahns f58a813a36 refactor(helm): disable the ingress per default
The default values forces people to create an ingress resources,
where per default an ingress is not necessary to start garage.

If someone wants to utilize an ingress, he would need to define
the values for the ingress either way, so enabling the ingress
explicitly makes more sense, then requiring it to be disabled per default
2022-12-11 23:11:56 +00:00
Alex defd7d9e63 Merge pull request 'Implement /health admin API endpoint to check node health' (#440) from admin-health-api into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/440
2022-12-11 17:25:28 +00:00
Alex Auvolat 533afcf4e1 simplify 2022-12-11 18:17:08 +01:00
Alex Auvolat 5ea5fd2130 Always return 200 OK on /v0/health, reinstate admin api doc as draft and complete it 2022-12-11 18:11:28 +01:00
Alex 35f8e8e2fb Merge pull request 'Fix typo in documentation' (#441) from felix.scheinost/garage:documentation-typo into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/441
2022-12-07 20:42:24 +00:00
felix.scheinost d5a2502b09 Fix typo in documentation 2022-12-07 12:43:49 +00:00
Alex Auvolat d7868c48a4 Separate /health (simple text answer) and /v0/health (full json answer, authenticated) 2022-12-05 15:38:32 +01:00
Alex Auvolat 280d1be7b1 Refactor health check and add ability to return it in json 2022-12-05 15:28:57 +01:00
Alex Auvolat 2065f011ca Implement /health admin API endpoint to check node health 2022-12-05 14:59:15 +01:00
Alex 243b7c9a1c Merge pull request 'Fix spelling mistake in docs' (#438) from tompearson/garage:fix-typo into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/438
2022-12-05 12:27:14 +00:00
tompearson a3afc761b6 Update 'doc/book/design/goals.md' 2022-12-04 16:27:46 +00:00
Alex 19bdd1c799 Merge pull request 'Fix logs appearing twice' (#435) from fix-logs into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/435
2022-11-29 21:30:39 +00:00
Alex 448dcc5cf4 Merge pull request 'Make repository into a Nix flake' (#424) from nix-remove-system into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/424
2022-11-29 21:26:41 +00:00
Alex Auvolat 26121bb619 Fix logs appearing twice 2022-11-29 22:23:27 +01:00
Alex 280330ac72 Merge pull request 'Add talk to the Capitole du Libre 2022' (#434) from CdL_talk into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/434
2022-11-27 13:38:13 +00:00
vincent 4d7b4d9d20 Add talk to the Capitole du Libre 2022 2022-11-27 11:36:01 +01:00
Alex fc450ec13a Merge pull request 'Fix #432: documentation issue' (#433) from fix-432 into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/433
2022-11-24 14:36:53 +00:00
Alex Auvolat 379b2049f5 Fix #432: documentation issue 2022-11-24 15:33:33 +01:00
Alex 293139a94a Merge pull request 'Tentative fix #414' (#429) from try-fix-414 into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/429
2022-11-21 21:45:17 +00:00
Alex Auvolat 54e800ef8d Tentative fix for issue #414 2022-11-21 17:13:41 +01:00
Alex Auvolat 60c26fbc62 Inject last modified date as git_version; flake cache uploading 2022-11-16 23:47:10 +01:00
Alex Auvolat e76dba9561 Make repository into a Nix flake 2022-11-16 23:25:34 +01:00
128 changed files with 4632 additions and 2195 deletions
-1
View File
@@ -1 +0,0 @@
*.pdf filter=lfs diff=lfs merge=lfs -text
Generated
+11 -14
View File
@@ -1048,7 +1048,7 @@ dependencies = [
[[package]]
name = "garage"
version = "0.8.0"
version = "0.8.1"
dependencies = [
"assert-json-diff",
"async-trait",
@@ -1080,7 +1080,6 @@ dependencies = [
"parse_duration",
"prometheus",
"rand 0.8.5",
"rmp-serde",
"serde",
"serde_bytes",
"serde_json",
@@ -1096,7 +1095,7 @@ dependencies = [
[[package]]
name = "garage_api"
version = "0.8.0"
version = "0.8.1"
dependencies = [
"async-trait",
"base64",
@@ -1141,7 +1140,7 @@ dependencies = [
[[package]]
name = "garage_block"
version = "0.8.0"
version = "0.8.1"
dependencies = [
"arc-swap",
"async-compression",
@@ -1156,7 +1155,6 @@ dependencies = [
"hex",
"opentelemetry",
"rand 0.8.5",
"rmp-serde",
"serde",
"serde_bytes",
"tokio",
@@ -1167,7 +1165,7 @@ dependencies = [
[[package]]
name = "garage_db"
version = "0.8.0"
version = "0.8.1"
dependencies = [
"clap 3.1.18",
"err-derive",
@@ -1182,7 +1180,7 @@ dependencies = [
[[package]]
name = "garage_model"
version = "0.8.0"
version = "0.8.1"
dependencies = [
"arc-swap",
"async-trait",
@@ -1200,7 +1198,6 @@ dependencies = [
"netapp",
"opentelemetry",
"rand 0.8.5",
"rmp-serde",
"serde",
"serde_bytes",
"tokio",
@@ -1210,7 +1207,7 @@ dependencies = [
[[package]]
name = "garage_rpc"
version = "0.8.0"
version = "0.8.1"
dependencies = [
"arc-swap",
"async-trait",
@@ -1229,7 +1226,6 @@ dependencies = [
"pnet_datalink",
"rand 0.8.5",
"reqwest",
"rmp-serde",
"schemars",
"serde",
"serde_bytes",
@@ -1241,8 +1237,9 @@ dependencies = [
[[package]]
name = "garage_table"
version = "0.8.0"
version = "0.8.1"
dependencies = [
"arc-swap",
"async-trait",
"bytes",
"futures",
@@ -1254,7 +1251,6 @@ dependencies = [
"hexdump",
"opentelemetry",
"rand 0.8.5",
"rmp-serde",
"serde",
"serde_bytes",
"tokio",
@@ -1263,7 +1259,7 @@ dependencies = [
[[package]]
name = "garage_util"
version = "0.8.0"
version = "0.8.1"
dependencies = [
"arc-swap",
"async-trait",
@@ -1276,6 +1272,7 @@ dependencies = [
"garage_db",
"git-version",
"hex",
"hexdump",
"http",
"hyper",
"lazy_static",
@@ -1294,7 +1291,7 @@ dependencies = [
[[package]]
name = "garage_web"
version = "0.8.0"
version = "0.8.1"
dependencies = [
"err-derive",
"futures",
+66 -70
View File
@@ -32,7 +32,7 @@ args@{
ignoreLockHash,
}:
let
nixifiedLockHash = "90b29705f5037c7e1b33f4650841f1266f2e86fa03d5d0c87ad80be7619985c7";
nixifiedLockHash = "b6aeefc112eb232904b24398f4e5da776c8ee2c13d427a26dbdf1732205d4fc9";
workspaceSrc = if args.workspaceSrc == null then ./. else args.workspaceSrc;
currentLockHash = builtins.hashFile "sha256" (workspaceSrc + /Cargo.lock);
lockHashIgnored = if ignoreLockHash
@@ -56,15 +56,15 @@ in
{
cargo2nixVersion = "0.11.0";
workspace = {
garage_db = rustPackages.unknown.garage_db."0.8.0";
garage_util = rustPackages.unknown.garage_util."0.8.0";
garage_rpc = rustPackages.unknown.garage_rpc."0.8.0";
garage_table = rustPackages.unknown.garage_table."0.8.0";
garage_block = rustPackages.unknown.garage_block."0.8.0";
garage_model = rustPackages.unknown.garage_model."0.8.0";
garage_api = rustPackages.unknown.garage_api."0.8.0";
garage_web = rustPackages.unknown.garage_web."0.8.0";
garage = rustPackages.unknown.garage."0.8.0";
garage_db = rustPackages.unknown.garage_db."0.8.1";
garage_util = rustPackages.unknown.garage_util."0.8.1";
garage_rpc = rustPackages.unknown.garage_rpc."0.8.1";
garage_table = rustPackages.unknown.garage_table."0.8.1";
garage_block = rustPackages.unknown.garage_block."0.8.1";
garage_model = rustPackages.unknown.garage_model."0.8.1";
garage_api = rustPackages.unknown.garage_api."0.8.1";
garage_web = rustPackages.unknown.garage_web."0.8.1";
garage = rustPackages.unknown.garage."0.8.1";
k2v-client = rustPackages.unknown.k2v-client."0.0.1";
};
"registry+https://github.com/rust-lang/crates.io-index".addr2line."0.17.0" = overridableMkRustCrate (profileName: rec {
@@ -1494,16 +1494,16 @@ in
};
});
"unknown".garage."0.8.0" = overridableMkRustCrate (profileName: rec {
"unknown".garage."0.8.1" = overridableMkRustCrate (profileName: rec {
name = "garage";
version = "0.8.0";
version = "0.8.1";
registry = "unknown";
src = fetchCrateLocal (workspaceSrc + "/src/garage");
features = builtins.concatLists [
(lib.optional (rootFeatures' ? "garage/bundled-libs" || rootFeatures' ? "garage/default") "bundled-libs")
(lib.optional (rootFeatures' ? "garage/consul-discovery") "consul-discovery")
(lib.optional (rootFeatures' ? "garage/default") "default")
(lib.optional (rootFeatures' ? "garage/k2v") "k2v")
(lib.optional (rootFeatures' ? "garage/default" || rootFeatures' ? "garage/k2v") "k2v")
(lib.optional (rootFeatures' ? "garage/kubernetes-discovery") "kubernetes-discovery")
(lib.optional (rootFeatures' ? "garage/lmdb") "lmdb")
(lib.optional (rootFeatures' ? "garage/default" || rootFeatures' ? "garage/metrics") "metrics")
@@ -1522,14 +1522,14 @@ in
bytesize = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".bytesize."1.1.0" { inherit profileName; }).out;
futures = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".futures."0.3.21" { inherit profileName; }).out;
futures_util = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".futures-util."0.3.21" { inherit profileName; }).out;
garage_api = (rustPackages."unknown".garage_api."0.8.0" { inherit profileName; }).out;
garage_block = (rustPackages."unknown".garage_block."0.8.0" { inherit profileName; }).out;
garage_db = (rustPackages."unknown".garage_db."0.8.0" { inherit profileName; }).out;
garage_model = (rustPackages."unknown".garage_model."0.8.0" { inherit profileName; }).out;
garage_rpc = (rustPackages."unknown".garage_rpc."0.8.0" { inherit profileName; }).out;
garage_table = (rustPackages."unknown".garage_table."0.8.0" { inherit profileName; }).out;
garage_util = (rustPackages."unknown".garage_util."0.8.0" { inherit profileName; }).out;
garage_web = (rustPackages."unknown".garage_web."0.8.0" { inherit profileName; }).out;
garage_api = (rustPackages."unknown".garage_api."0.8.1" { inherit profileName; }).out;
garage_block = (rustPackages."unknown".garage_block."0.8.1" { inherit profileName; }).out;
garage_db = (rustPackages."unknown".garage_db."0.8.1" { inherit profileName; }).out;
garage_model = (rustPackages."unknown".garage_model."0.8.1" { inherit profileName; }).out;
garage_rpc = (rustPackages."unknown".garage_rpc."0.8.1" { inherit profileName; }).out;
garage_table = (rustPackages."unknown".garage_table."0.8.1" { inherit profileName; }).out;
garage_util = (rustPackages."unknown".garage_util."0.8.1" { inherit profileName; }).out;
garage_web = (rustPackages."unknown".garage_web."0.8.1" { inherit profileName; }).out;
hex = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".hex."0.4.3" { inherit profileName; }).out;
sodiumoxide = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".kuska-sodiumoxide."0.2.5-0" { inherit profileName; }).out;
netapp = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".netapp."0.5.2" { inherit profileName; }).out;
@@ -1539,7 +1539,6 @@ in
parse_duration = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".parse_duration."2.1.1" { inherit profileName; }).out;
${ if rootFeatures' ? "garage/default" || rootFeatures' ? "garage/metrics" || rootFeatures' ? "garage/prometheus" then "prometheus" else null } = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".prometheus."0.13.0" { inherit profileName; }).out;
rand = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".rand."0.8.5" { inherit profileName; }).out;
rmp_serde = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".rmp-serde."0.15.5" { inherit profileName; }).out;
serde = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".serde."1.0.137" { inherit profileName; }).out;
serde_bytes = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".serde_bytes."0.11.5" { inherit profileName; }).out;
structopt = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".structopt."0.3.26" { inherit profileName; }).out;
@@ -1563,13 +1562,13 @@ in
};
});
"unknown".garage_api."0.8.0" = overridableMkRustCrate (profileName: rec {
"unknown".garage_api."0.8.1" = overridableMkRustCrate (profileName: rec {
name = "garage_api";
version = "0.8.0";
version = "0.8.1";
registry = "unknown";
src = fetchCrateLocal (workspaceSrc + "/src/api");
features = builtins.concatLists [
(lib.optional (rootFeatures' ? "garage/k2v" || rootFeatures' ? "garage_api/k2v") "k2v")
(lib.optional (rootFeatures' ? "garage/default" || rootFeatures' ? "garage/k2v" || rootFeatures' ? "garage_api/k2v") "k2v")
(lib.optional (rootFeatures' ? "garage/default" || rootFeatures' ? "garage/metrics" || rootFeatures' ? "garage_api/metrics") "metrics")
(lib.optional (rootFeatures' ? "garage/default" || rootFeatures' ? "garage/metrics" || rootFeatures' ? "garage_api/metrics" || rootFeatures' ? "garage_api/opentelemetry-prometheus") "opentelemetry-prometheus")
(lib.optional (rootFeatures' ? "garage/default" || rootFeatures' ? "garage/metrics" || rootFeatures' ? "garage_api/metrics" || rootFeatures' ? "garage_api/prometheus") "prometheus")
@@ -1584,11 +1583,11 @@ in
form_urlencoded = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".form_urlencoded."1.0.1" { inherit profileName; }).out;
futures = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".futures."0.3.21" { inherit profileName; }).out;
futures_util = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".futures-util."0.3.21" { inherit profileName; }).out;
garage_block = (rustPackages."unknown".garage_block."0.8.0" { inherit profileName; }).out;
garage_model = (rustPackages."unknown".garage_model."0.8.0" { inherit profileName; }).out;
garage_rpc = (rustPackages."unknown".garage_rpc."0.8.0" { inherit profileName; }).out;
garage_table = (rustPackages."unknown".garage_table."0.8.0" { inherit profileName; }).out;
garage_util = (rustPackages."unknown".garage_util."0.8.0" { inherit profileName; }).out;
garage_block = (rustPackages."unknown".garage_block."0.8.1" { inherit profileName; }).out;
garage_model = (rustPackages."unknown".garage_model."0.8.1" { inherit profileName; }).out;
garage_rpc = (rustPackages."unknown".garage_rpc."0.8.1" { inherit profileName; }).out;
garage_table = (rustPackages."unknown".garage_table."0.8.1" { inherit profileName; }).out;
garage_util = (rustPackages."unknown".garage_util."0.8.1" { inherit profileName; }).out;
hex = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".hex."0.4.3" { inherit profileName; }).out;
hmac = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".hmac."0.12.1" { inherit profileName; }).out;
http = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".http."0.2.8" { inherit profileName; }).out;
@@ -1617,9 +1616,9 @@ in
};
});
"unknown".garage_block."0.8.0" = overridableMkRustCrate (profileName: rec {
"unknown".garage_block."0.8.1" = overridableMkRustCrate (profileName: rec {
name = "garage_block";
version = "0.8.0";
version = "0.8.1";
registry = "unknown";
src = fetchCrateLocal (workspaceSrc + "/src/block");
features = builtins.concatLists [
@@ -1632,14 +1631,13 @@ in
bytes = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".bytes."1.2.0" { inherit profileName; }).out;
futures = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".futures."0.3.21" { inherit profileName; }).out;
futures_util = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".futures-util."0.3.21" { inherit profileName; }).out;
garage_db = (rustPackages."unknown".garage_db."0.8.0" { inherit profileName; }).out;
garage_rpc = (rustPackages."unknown".garage_rpc."0.8.0" { inherit profileName; }).out;
garage_table = (rustPackages."unknown".garage_table."0.8.0" { inherit profileName; }).out;
garage_util = (rustPackages."unknown".garage_util."0.8.0" { inherit profileName; }).out;
garage_db = (rustPackages."unknown".garage_db."0.8.1" { inherit profileName; }).out;
garage_rpc = (rustPackages."unknown".garage_rpc."0.8.1" { inherit profileName; }).out;
garage_table = (rustPackages."unknown".garage_table."0.8.1" { inherit profileName; }).out;
garage_util = (rustPackages."unknown".garage_util."0.8.1" { inherit profileName; }).out;
hex = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".hex."0.4.3" { inherit profileName; }).out;
opentelemetry = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".opentelemetry."0.17.0" { inherit profileName; }).out;
rand = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".rand."0.8.5" { inherit profileName; }).out;
rmp_serde = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".rmp-serde."0.15.5" { inherit profileName; }).out;
serde = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".serde."1.0.137" { inherit profileName; }).out;
serde_bytes = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".serde_bytes."0.11.5" { inherit profileName; }).out;
tokio = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".tokio."1.17.0" { inherit profileName; }).out;
@@ -1649,9 +1647,9 @@ in
};
});
"unknown".garage_db."0.8.0" = overridableMkRustCrate (profileName: rec {
"unknown".garage_db."0.8.1" = overridableMkRustCrate (profileName: rec {
name = "garage_db";
version = "0.8.0";
version = "0.8.1";
registry = "unknown";
src = fetchCrateLocal (workspaceSrc + "/src/db");
features = builtins.concatLists [
@@ -1681,14 +1679,14 @@ in
};
});
"unknown".garage_model."0.8.0" = overridableMkRustCrate (profileName: rec {
"unknown".garage_model."0.8.1" = overridableMkRustCrate (profileName: rec {
name = "garage_model";
version = "0.8.0";
version = "0.8.1";
registry = "unknown";
src = fetchCrateLocal (workspaceSrc + "/src/model");
features = builtins.concatLists [
[ "default" ]
(lib.optional (rootFeatures' ? "garage/k2v" || rootFeatures' ? "garage_api/k2v" || rootFeatures' ? "garage_model/k2v") "k2v")
(lib.optional (rootFeatures' ? "garage/default" || rootFeatures' ? "garage/k2v" || rootFeatures' ? "garage_api/k2v" || rootFeatures' ? "garage_model/k2v") "k2v")
(lib.optional (rootFeatures' ? "garage/lmdb" || rootFeatures' ? "garage_model/lmdb") "lmdb")
[ "sled" ]
(lib.optional (rootFeatures' ? "garage/sqlite" || rootFeatures' ? "garage_model/sqlite") "sqlite")
@@ -1701,16 +1699,15 @@ in
err_derive = (buildRustPackages."registry+https://github.com/rust-lang/crates.io-index".err-derive."0.3.1" { profileName = "__noProfile"; }).out;
futures = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".futures."0.3.21" { inherit profileName; }).out;
futures_util = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".futures-util."0.3.21" { inherit profileName; }).out;
garage_block = (rustPackages."unknown".garage_block."0.8.0" { inherit profileName; }).out;
garage_db = (rustPackages."unknown".garage_db."0.8.0" { inherit profileName; }).out;
garage_rpc = (rustPackages."unknown".garage_rpc."0.8.0" { inherit profileName; }).out;
garage_table = (rustPackages."unknown".garage_table."0.8.0" { inherit profileName; }).out;
garage_util = (rustPackages."unknown".garage_util."0.8.0" { inherit profileName; }).out;
garage_block = (rustPackages."unknown".garage_block."0.8.1" { inherit profileName; }).out;
garage_db = (rustPackages."unknown".garage_db."0.8.1" { inherit profileName; }).out;
garage_rpc = (rustPackages."unknown".garage_rpc."0.8.1" { inherit profileName; }).out;
garage_table = (rustPackages."unknown".garage_table."0.8.1" { inherit profileName; }).out;
garage_util = (rustPackages."unknown".garage_util."0.8.1" { inherit profileName; }).out;
hex = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".hex."0.4.3" { inherit profileName; }).out;
netapp = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".netapp."0.5.2" { inherit profileName; }).out;
opentelemetry = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".opentelemetry."0.17.0" { inherit profileName; }).out;
rand = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".rand."0.8.5" { inherit profileName; }).out;
rmp_serde = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".rmp-serde."0.15.5" { inherit profileName; }).out;
serde = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".serde."1.0.137" { inherit profileName; }).out;
serde_bytes = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".serde_bytes."0.11.5" { inherit profileName; }).out;
tokio = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".tokio."1.17.0" { inherit profileName; }).out;
@@ -1719,9 +1716,9 @@ in
};
});
"unknown".garage_rpc."0.8.0" = overridableMkRustCrate (profileName: rec {
"unknown".garage_rpc."0.8.1" = overridableMkRustCrate (profileName: rec {
name = "garage_rpc";
version = "0.8.0";
version = "0.8.1";
registry = "unknown";
src = fetchCrateLocal (workspaceSrc + "/src/rpc");
features = builtins.concatLists [
@@ -1741,7 +1738,7 @@ in
${ if rootFeatures' ? "garage/consul-discovery" || rootFeatures' ? "garage_rpc/consul-discovery" || rootFeatures' ? "garage_rpc/err-derive" then "err_derive" else null } = (buildRustPackages."registry+https://github.com/rust-lang/crates.io-index".err-derive."0.3.1" { profileName = "__noProfile"; }).out;
futures = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".futures."0.3.21" { inherit profileName; }).out;
futures_util = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".futures-util."0.3.21" { inherit profileName; }).out;
garage_util = (rustPackages."unknown".garage_util."0.8.0" { inherit profileName; }).out;
garage_util = (rustPackages."unknown".garage_util."0.8.1" { inherit profileName; }).out;
gethostname = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".gethostname."0.2.3" { inherit profileName; }).out;
hex = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".hex."0.4.3" { inherit profileName; }).out;
${ if rootFeatures' ? "garage/kubernetes-discovery" || rootFeatures' ? "garage_rpc/k8s-openapi" || rootFeatures' ? "garage_rpc/kubernetes-discovery" then "k8s_openapi" else null } = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".k8s-openapi."0.16.0" { inherit profileName; }).out;
@@ -1752,7 +1749,6 @@ in
pnet_datalink = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".pnet_datalink."0.28.0" { inherit profileName; }).out;
rand = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".rand."0.8.5" { inherit profileName; }).out;
${ if rootFeatures' ? "garage/consul-discovery" || rootFeatures' ? "garage_rpc/consul-discovery" || rootFeatures' ? "garage_rpc/reqwest" then "reqwest" else null } = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".reqwest."0.11.12" { inherit profileName; }).out;
rmp_serde = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".rmp-serde."0.15.5" { inherit profileName; }).out;
${ if rootFeatures' ? "garage/kubernetes-discovery" || rootFeatures' ? "garage_rpc/kubernetes-discovery" || rootFeatures' ? "garage_rpc/schemars" then "schemars" else null } = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".schemars."0.8.8" { inherit profileName; }).out;
serde = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".serde."1.0.137" { inherit profileName; }).out;
serde_bytes = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".serde_bytes."0.11.5" { inherit profileName; }).out;
@@ -1763,24 +1759,24 @@ in
};
});
"unknown".garage_table."0.8.0" = overridableMkRustCrate (profileName: rec {
"unknown".garage_table."0.8.1" = overridableMkRustCrate (profileName: rec {
name = "garage_table";
version = "0.8.0";
version = "0.8.1";
registry = "unknown";
src = fetchCrateLocal (workspaceSrc + "/src/table");
dependencies = {
arc_swap = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".arc-swap."1.5.0" { inherit profileName; }).out;
async_trait = (buildRustPackages."registry+https://github.com/rust-lang/crates.io-index".async-trait."0.1.52" { profileName = "__noProfile"; }).out;
bytes = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".bytes."1.2.0" { inherit profileName; }).out;
futures = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".futures."0.3.21" { inherit profileName; }).out;
futures_util = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".futures-util."0.3.21" { inherit profileName; }).out;
garage_db = (rustPackages."unknown".garage_db."0.8.0" { inherit profileName; }).out;
garage_rpc = (rustPackages."unknown".garage_rpc."0.8.0" { inherit profileName; }).out;
garage_util = (rustPackages."unknown".garage_util."0.8.0" { inherit profileName; }).out;
garage_db = (rustPackages."unknown".garage_db."0.8.1" { inherit profileName; }).out;
garage_rpc = (rustPackages."unknown".garage_rpc."0.8.1" { inherit profileName; }).out;
garage_util = (rustPackages."unknown".garage_util."0.8.1" { inherit profileName; }).out;
hex = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".hex."0.4.3" { inherit profileName; }).out;
hexdump = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".hexdump."0.1.1" { inherit profileName; }).out;
opentelemetry = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".opentelemetry."0.17.0" { inherit profileName; }).out;
rand = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".rand."0.8.5" { inherit profileName; }).out;
rmp_serde = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".rmp-serde."0.15.5" { inherit profileName; }).out;
serde = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".serde."1.0.137" { inherit profileName; }).out;
serde_bytes = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".serde_bytes."0.11.5" { inherit profileName; }).out;
tokio = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".tokio."1.17.0" { inherit profileName; }).out;
@@ -1788,13 +1784,13 @@ in
};
});
"unknown".garage_util."0.8.0" = overridableMkRustCrate (profileName: rec {
"unknown".garage_util."0.8.1" = overridableMkRustCrate (profileName: rec {
name = "garage_util";
version = "0.8.0";
version = "0.8.1";
registry = "unknown";
src = fetchCrateLocal (workspaceSrc + "/src/util");
features = builtins.concatLists [
(lib.optional (rootFeatures' ? "garage/k2v" || rootFeatures' ? "garage_api/k2v" || rootFeatures' ? "garage_model/k2v" || rootFeatures' ? "garage_util/k2v") "k2v")
(lib.optional (rootFeatures' ? "garage/default" || rootFeatures' ? "garage/k2v" || rootFeatures' ? "garage_api/k2v" || rootFeatures' ? "garage_model/k2v" || rootFeatures' ? "garage_util/k2v") "k2v")
];
dependencies = {
arc_swap = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".arc-swap."1.5.0" { inherit profileName; }).out;
@@ -1805,9 +1801,10 @@ in
digest = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".digest."0.10.3" { inherit profileName; }).out;
err_derive = (buildRustPackages."registry+https://github.com/rust-lang/crates.io-index".err-derive."0.3.1" { profileName = "__noProfile"; }).out;
futures = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".futures."0.3.21" { inherit profileName; }).out;
garage_db = (rustPackages."unknown".garage_db."0.8.0" { inherit profileName; }).out;
garage_db = (rustPackages."unknown".garage_db."0.8.1" { inherit profileName; }).out;
git_version = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".git-version."0.3.5" { inherit profileName; }).out;
hex = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".hex."0.4.3" { inherit profileName; }).out;
hexdump = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".hexdump."0.1.1" { inherit profileName; }).out;
http = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".http."0.2.8" { inherit profileName; }).out;
hyper = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".hyper."0.14.18" { inherit profileName; }).out;
lazy_static = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".lazy_static."1.4.0" { inherit profileName; }).out;
@@ -1825,18 +1822,18 @@ in
};
});
"unknown".garage_web."0.8.0" = overridableMkRustCrate (profileName: rec {
"unknown".garage_web."0.8.1" = overridableMkRustCrate (profileName: rec {
name = "garage_web";
version = "0.8.0";
version = "0.8.1";
registry = "unknown";
src = fetchCrateLocal (workspaceSrc + "/src/web");
dependencies = {
err_derive = (buildRustPackages."registry+https://github.com/rust-lang/crates.io-index".err-derive."0.3.1" { profileName = "__noProfile"; }).out;
futures = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".futures."0.3.21" { inherit profileName; }).out;
garage_api = (rustPackages."unknown".garage_api."0.8.0" { inherit profileName; }).out;
garage_model = (rustPackages."unknown".garage_model."0.8.0" { inherit profileName; }).out;
garage_table = (rustPackages."unknown".garage_table."0.8.0" { inherit profileName; }).out;
garage_util = (rustPackages."unknown".garage_util."0.8.0" { inherit profileName; }).out;
garage_api = (rustPackages."unknown".garage_api."0.8.1" { inherit profileName; }).out;
garage_model = (rustPackages."unknown".garage_model."0.8.1" { inherit profileName; }).out;
garage_table = (rustPackages."unknown".garage_table."0.8.1" { inherit profileName; }).out;
garage_util = (rustPackages."unknown".garage_util."0.8.1" { inherit profileName; }).out;
http = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".http."0.2.8" { inherit profileName; }).out;
hyper = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".hyper."0.14.18" { inherit profileName; }).out;
opentelemetry = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".opentelemetry."0.17.0" { inherit profileName; }).out;
@@ -2450,7 +2447,7 @@ in
dependencies = {
base64 = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".base64."0.13.0" { inherit profileName; }).out;
${ if rootFeatures' ? "k2v-client/clap" || rootFeatures' ? "k2v-client/cli" then "clap" else null } = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".clap."3.1.18" { inherit profileName; }).out;
${ if rootFeatures' ? "k2v-client/cli" || rootFeatures' ? "k2v-client/garage_util" then "garage_util" else null } = (rustPackages."unknown".garage_util."0.8.0" { inherit profileName; }).out;
${ if rootFeatures' ? "k2v-client/cli" || rootFeatures' ? "k2v-client/garage_util" then "garage_util" else null } = (rustPackages."unknown".garage_util."0.8.1" { inherit profileName; }).out;
http = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".http."0.2.8" { inherit profileName; }).out;
log = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".log."0.4.16" { inherit profileName; }).out;
rusoto_core = (rustPackages."registry+https://github.com/rust-lang/crates.io-index".rusoto_core."0.48.0" { inherit profileName; }).out;
@@ -5398,7 +5395,6 @@ in
[ "attributes" ]
[ "default" ]
[ "log" ]
[ "log-always" ]
[ "std" ]
[ "tracing-attributes" ]
];
+4 -4
View File
@@ -11,14 +11,14 @@ let
build_debug_and_release = (target: {
debug = (compile {
inherit target git_version;
inherit system target git_version pkgsSrc cargo2nixOverlay;
release = false;
}).workspace.garage {
compileMode = "build";
};
release = (compile {
inherit target git_version;
inherit system target git_version pkgsSrc cargo2nixOverlay;
release = true;
}).workspace.garage {
compileMode = "build";
@@ -39,7 +39,7 @@ in {
};
test = {
amd64 = test (compile {
inherit git_version;
inherit system git_version pkgsSrc cargo2nixOverlay;
target = "x86_64-unknown-linux-musl";
features = [
"garage/bundled-libs"
@@ -52,7 +52,7 @@ in {
};
clippy = {
amd64 = (compile {
inherit git_version;
inherit system git_version pkgsSrc cargo2nixOverlay;
target = "x86_64-unknown-linux-musl";
compiler = "clippy";
}).workspace.garage {
+51 -8
View File
@@ -5,16 +5,59 @@ weight = 20
## S3
### Using Minio SDK
First install the SDK:
```bash
pip3 install minio
```
Then instantiate a client object using garage root domain, api key and secret:
```python
import minio
client = minio.Minio(
"your.domain.tld",
"GKyourapikey",
"abcd[...]1234",
# Force the region, this is specific to garage
region="region",
)
```
Then use all the standard S3 endpoints as implemented by the Minio SDK:
```
# List buckets
print(client.list_buckets())
# Put an object containing 'content' to /path in bucket named 'bucket':
content = b"content"
client.put_object(
"bucket",
"path",
io.BytesIO(content),
len(content),
)
# Read the object back and check contents
data = client.get_object("bucket", "path").read()
assert data == content
```
For further documentation, see the Minio SDK
[Reference](https://docs.min.io/docs/python-client-api-reference.html)
### Using Amazon boto3
*Coming soon*
Some refs:
- Minio SDK
- [Reference](https://docs.min.io/docs/python-client-api-reference.html)
- Amazon boto3
- [Installation](https://boto3.amazonaws.com/v1/documentation/api/latest/guide/quickstart.html)
- [Reference](https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/s3.html)
- [Example](https://boto3.amazonaws.com/v1/documentation/api/latest/guide/s3-uploading-files.html)
See the official documentation:
- [Installation](https://boto3.amazonaws.com/v1/documentation/api/latest/guide/quickstart.html)
- [Reference](https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/s3.html)
- [Example](https://boto3.amazonaws.com/v1/documentation/api/latest/guide/s3-uploading-files.html)
## K2V
+10 -1
View File
@@ -8,7 +8,7 @@ In this section, we cover the following web applications:
| Name | Status | Note |
|------|--------|------|
| [Nextcloud](#nextcloud) | ✅ | Both Primary Storage and External Storage are supported |
| [Peertube](#peertube) | ✅ | Must be configured with the website endpoint |
| [Peertube](#peertube) | ✅ | Supported with the website endpoint, proxifying private videos unsupported |
| [Mastodon](#mastodon) | ✅ | Natively supported |
| [Matrix](#matrix) | ✅ | Tested with `synapse-s3-storage-provider` |
| [Pixelfed](#pixelfed) | ❓ | Not yet tested |
@@ -128,6 +128,10 @@ In other words, Peertube is only responsible of the "control plane" and offload
In return, this system is a bit harder to configure.
We show how it is still possible to configure Garage with Peertube, allowing you to spread the load and the bandwidth usage on the Garage cluster.
Starting from version 5.0, Peertube also supports improving the security for private videos by not exposing them directly
but relying on a single control point in the Peertube instance. This is based on S3 per-object and prefix ACL, which are not currently supported
in Garage, so this feature is unsupported. While this technically impedes security for private videos, it is not a blocking issue and could be
a reasonable trade-off for some instances.
### Create resources in Garage
@@ -195,6 +199,11 @@ object_storage:
max_upload_part: 2GB
proxy:
# You may enable this feature, yet it will not provide any security benefit, so
# you should rather benefit from Garage public endpoint for all videos
proxify_private_files: false
streaming_playlists:
bucket_name: 'peertube-playlist'
+1 -1
View File
@@ -109,7 +109,7 @@ especially you must consider the following folders/files:
this folder will be your main data storage and must be on a large storage (e.g. large HDD)
A valid `/etc/garage/garage.toml` for our cluster would look as follows:
A valid `/etc/garage.toml` for our cluster would look as follows:
```toml
metadata_dir = "/var/lib/garage/meta"
+1 -1
View File
@@ -12,7 +12,7 @@ as pictures, video, images, documents, etc., in a redundant multi-node
setting. S3 is versatile enough to also be used to publish a static
website.
Garage is an opinionated object storage solutoin, we focus on the following **desirable properties**:
Garage is an opinionated object storage solution, we focus on the following **desirable properties**:
- **Internet enabled**: made for multi-sites (eg. datacenters, offices, households, etc.) interconnected through regular Internet connections.
- **Self-contained & lightweight**: works everywhere and integrates well in existing environments to target [hyperconverged infrastructures](https://en.wikipedia.org/wiki/Hyper-converged_infrastructure).
+1 -1
View File
@@ -83,7 +83,7 @@ This feature is totally invisible to S3 clients and does not break compatibility
### Cluster administration API
Garage provides a fully-fledged REST API to administer your cluster programatically.
Functionnality included in the admin API include: setting up and monitoring
Functionality included in the admin API include: setting up and monitoring
cluster nodes, managing access credentials, and managing storage buckets and bucket aliases.
A full reference of the administration API is available [here](@/documentation/reference-manual/admin-api.md).
+686
View File
@@ -0,0 +1,686 @@
+++
title = "Administration API"
weight = 60
+++
The Garage administration API is accessible through a dedicated server whose
listen address is specified in the `[admin]` section of the configuration
file (see [configuration file
reference](@/documentation/reference-manual/configuration.md))
**WARNING.** At this point, there is no comittement to stability of the APIs described in this document.
We will bump the version numbers prefixed to each API endpoint at each time the syntax
or semantics change, meaning that code that relies on these endpoint will break
when changes are introduced.
The Garage administration API was introduced in version 0.7.2, this document
does not apply to older versions of Garage.
## Access control
The admin API uses two different tokens for acces control, that are specified in the config file's `[admin]` section:
- `metrics_token`: the token for accessing the Metrics endpoint (if this token
is not set in the config file, the Metrics endpoint can be accessed without
access control);
- `admin_token`: the token for accessing all of the other administration
endpoints (if this token is not set in the config file, access to these
endpoints is disabled entirely).
These tokens are used as simple HTTP bearer tokens. In other words, to
authenticate access to an admin API endpoint, add the following HTTP header
to your request:
```
Authorization: Bearer <token>
```
## Administration API endpoints
### Metrics-related endpoints
#### Metrics `GET /metrics`
Returns internal Garage metrics in Prometheus format.
#### Health `GET /health`
Used for simple health checks in a cluster setting with an orchestrator.
Returns an HTTP status 200 if the node is ready to answer user's requests,
and an HTTP status 503 (Service Unavailable) if there are some partitions
for which a quorum of nodes is not available.
A simple textual message is also returned in a body with content-type `text/plain`.
See `/v0/health` for an API that also returns JSON output.
### Cluster operations
#### GetClusterStatus `GET /v0/status`
Returns the cluster's current status in JSON, including:
- ID of the node being queried and its version of the Garage daemon
- Live nodes
- Currently configured cluster layout
- Staged changes to the cluster layout
Example response body:
```json
{
"node": "ec79480e0ce52ae26fd00c9da684e4fa56658d9c64cdcecb094e936de0bfe71f",
"garage_version": "git:v0.8.0",
"knownNodes": {
"ec79480e0ce52ae26fd00c9da684e4fa56658d9c64cdcecb094e936de0bfe71f": {
"addr": "10.0.0.11:3901",
"is_up": true,
"last_seen_secs_ago": 9,
"hostname": "node1"
},
"4a6ae5a1d0d33bf895f5bb4f0a418b7dc94c47c0dd2eb108d1158f3c8f60b0ff": {
"addr": "10.0.0.12:3901",
"is_up": true,
"last_seen_secs_ago": 1,
"hostname": "node2"
},
"23ffd0cdd375ebff573b20cc5cef38996b51c1a7d6dbcf2c6e619876e507cf27": {
"addr": "10.0.0.21:3901",
"is_up": true,
"last_seen_secs_ago": 7,
"hostname": "node3"
},
"e2ee7984ee65b260682086ec70026165903c86e601a4a5a501c1900afe28d84b": {
"addr": "10.0.0.22:3901",
"is_up": true,
"last_seen_secs_ago": 1,
"hostname": "node4"
}
},
"layout": {
"version": 12,
"roles": {
"ec79480e0ce52ae26fd00c9da684e4fa56658d9c64cdcecb094e936de0bfe71f": {
"zone": "dc1",
"capacity": 4,
"tags": [
"node1"
]
},
"4a6ae5a1d0d33bf895f5bb4f0a418b7dc94c47c0dd2eb108d1158f3c8f60b0ff": {
"zone": "dc1",
"capacity": 6,
"tags": [
"node2"
]
},
"23ffd0cdd375ebff573b20cc5cef38996b51c1a7d6dbcf2c6e619876e507cf27": {
"zone": "dc2",
"capacity": 10,
"tags": [
"node3"
]
}
},
"stagedRoleChanges": {
"e2ee7984ee65b260682086ec70026165903c86e601a4a5a501c1900afe28d84b": {
"zone": "dc2",
"capacity": 5,
"tags": [
"node4"
]
}
}
}
}
```
#### GetClusterHealth `GET /v0/health`
Returns the cluster's current health in JSON format, with the following variables:
- `status`: one of `Healthy`, `Degraded` or `Unavailable`:
- Healthy: Garage node is connected to all storage nodes
- Degraded: Garage node is not connected to all storage nodes, but a quorum of write nodes is available for all partitions
- Unavailable: a quorum of write nodes is not available for some partitions
- `known_nodes`: the number of nodes this Garage node has had a TCP connection to since the daemon started
- `connected_nodes`: the nubmer of nodes this Garage node currently has an open connection to
- `storage_nodes`: the number of storage nodes currently registered in the cluster layout
- `storage_nodes_ok`: the number of storage nodes to which a connection is currently open
- `partitions`: the total number of partitions of the data (currently always 256)
- `partitions_quorum`: the number of partitions for which a quorum of write nodes is available
- `partitions_all_ok`: the number of partitions for which we are connected to all storage nodes responsible of storing it
Contrarily to `GET /health`, this endpoint always returns a 200 OK HTTP response code.
Example response body:
```json
{
"status": "Degraded",
"known_nodes": 3,
"connected_nodes": 2,
"storage_nodes": 3,
"storage_nodes_ok": 2,
"partitions": 256,
"partitions_quorum": 256,
"partitions_all_ok": 0
}
```
#### ConnectClusterNodes `POST /v0/connect`
Instructs this Garage node to connect to other Garage nodes at specified addresses.
Example request body:
```json
[
"ec79480e0ce52ae26fd00c9da684e4fa56658d9c64cdcecb094e936de0bfe71f@10.0.0.11:3901",
"4a6ae5a1d0d33bf895f5bb4f0a418b7dc94c47c0dd2eb108d1158f3c8f60b0ff@10.0.0.12:3901"
]
```
The format of the string for a node to connect to is: `<node ID>@<ip address>:<port>`, same as in the `garage node connect` CLI call.
Example response:
```json
[
{
"success": true,
"error": null
},
{
"success": false,
"error": "Handshake error"
}
]
```
#### GetClusterLayout `GET /v0/layout`
Returns the cluster's current layout in JSON, including:
- Currently configured cluster layout
- Staged changes to the cluster layout
(the info returned by this endpoint is a subset of the info returned by GetClusterStatus)
Example response body:
```json
{
"version": 12,
"roles": {
"ec79480e0ce52ae26fd00c9da684e4fa56658d9c64cdcecb094e936de0bfe71f": {
"zone": "dc1",
"capacity": 4,
"tags": [
"node1"
]
},
"4a6ae5a1d0d33bf895f5bb4f0a418b7dc94c47c0dd2eb108d1158f3c8f60b0ff": {
"zone": "dc1",
"capacity": 6,
"tags": [
"node2"
]
},
"23ffd0cdd375ebff573b20cc5cef38996b51c1a7d6dbcf2c6e619876e507cf27": {
"zone": "dc2",
"capacity": 10,
"tags": [
"node3"
]
}
},
"stagedRoleChanges": {
"e2ee7984ee65b260682086ec70026165903c86e601a4a5a501c1900afe28d84b": {
"zone": "dc2",
"capacity": 5,
"tags": [
"node4"
]
}
}
}
```
#### UpdateClusterLayout `POST /v0/layout`
Send modifications to the cluster layout. These modifications will
be included in the staged role changes, visible in subsequent calls
of `GetClusterLayout`. Once the set of staged changes is satisfactory,
the user may call `ApplyClusterLayout` to apply the changed changes,
or `Revert ClusterLayout` to clear all of the staged changes in
the layout.
Request body format:
```json
{
<node_id>: {
"capacity": <new_capacity>,
"zone": <new_zone>,
"tags": [
<new_tag>,
...
]
},
<node_id_to_remove>: null,
...
}
```
Contrary to the CLI that may update only a subset of the fields
`capacity`, `zone` and `tags`, when calling this API all of these
values must be specified.
#### ApplyClusterLayout `POST /v0/layout/apply`
Applies to the cluster the layout changes currently registered as
staged layout changes.
Request body format:
```json
{
"version": 13
}
```
Similarly to the CLI, the body must include the version of the new layout
that will be created, which MUST be 1 + the value of the currently
existing layout in the cluster.
#### RevertClusterLayout `POST /v0/layout/revert`
Clears all of the staged layout changes.
Request body format:
```json
{
"version": 13
}
```
Reverting the staged changes is done by incrementing the version number
and clearing the contents of the staged change list.
Similarly to the CLI, the body must include the incremented
version number, which MUST be 1 + the value of the currently
existing layout in the cluster.
### Access key operations
#### ListKeys `GET /v0/key`
Returns all API access keys in the cluster.
Example response:
```json
[
{
"id": "GK31c2f218a2e44f485b94239e",
"name": "test"
},
{
"id": "GKe10061ac9c2921f09e4c5540",
"name": "test2"
}
]
```
#### CreateKey `POST /v0/key`
Creates a new API access key.
Request body format:
```json
{
"name": "NameOfMyKey"
}
```
#### ImportKey `POST /v0/key/import`
Imports an existing API key.
Request body format:
```json
{
"accessKeyId": "GK31c2f218a2e44f485b94239e",
"secretAccessKey": "b892c0665f0ada8a4755dae98baa3b133590e11dae3bcc1f9d769d67f16c3835",
"name": "NameOfMyKey"
}
```
#### GetKeyInfo `GET /v0/key?id=<acces key id>`
#### GetKeyInfo `GET /v0/key?search=<pattern>`
Returns information about the requested API access key.
If `id` is set, the key is looked up using its exact identifier (faster).
If `search` is set, the key is looked up using its name or prefix
of identifier (slower, all keys are enumerated to do this).
Example response:
```json
{
"name": "test",
"accessKeyId": "GK31c2f218a2e44f485b94239e",
"secretAccessKey": "b892c0665f0ada8a4755dae98baa3b133590e11dae3bcc1f9d769d67f16c3835",
"permissions": {
"createBucket": false
},
"buckets": [
{
"id": "70dc3bed7fe83a75e46b66e7ddef7d56e65f3c02f9f80b6749fb97eccb5e1033",
"globalAliases": [
"test2"
],
"localAliases": [],
"permissions": {
"read": true,
"write": true,
"owner": false
}
},
{
"id": "d7452a935e663fc1914f3a5515163a6d3724010ce8dfd9e4743ca8be5974f995",
"globalAliases": [
"test3"
],
"localAliases": [],
"permissions": {
"read": true,
"write": true,
"owner": false
}
},
{
"id": "e6a14cd6a27f48684579ec6b381c078ab11697e6bc8513b72b2f5307e25fff9b",
"globalAliases": [],
"localAliases": [
"test"
],
"permissions": {
"read": true,
"write": true,
"owner": true
}
},
{
"id": "96470e0df00ec28807138daf01915cfda2bee8eccc91dea9558c0b4855b5bf95",
"globalAliases": [
"alex"
],
"localAliases": [],
"permissions": {
"read": true,
"write": true,
"owner": true
}
}
]
}
```
#### DeleteKey `DELETE /v0/key?id=<acces key id>`
Deletes an API access key.
#### UpdateKey `POST /v0/key?id=<acces key id>`
Updates information about the specified API access key.
Request body format:
```json
{
"name": "NameOfMyKey",
"allow": {
"createBucket": true,
},
"deny": {}
}
```
All fields (`name`, `allow` and `deny`) are optionnal.
If they are present, the corresponding modifications are applied to the key, otherwise nothing is changed.
The possible flags in `allow` and `deny` are: `createBucket`.
### Bucket operations
#### ListBuckets `GET /v0/bucket`
Returns all storage buckets in the cluster.
Example response:
```json
[
{
"id": "70dc3bed7fe83a75e46b66e7ddef7d56e65f3c02f9f80b6749fb97eccb5e1033",
"globalAliases": [
"test2"
],
"localAliases": []
},
{
"id": "96470e0df00ec28807138daf01915cfda2bee8eccc91dea9558c0b4855b5bf95",
"globalAliases": [
"alex"
],
"localAliases": []
},
{
"id": "d7452a935e663fc1914f3a5515163a6d3724010ce8dfd9e4743ca8be5974f995",
"globalAliases": [
"test3"
],
"localAliases": []
},
{
"id": "e6a14cd6a27f48684579ec6b381c078ab11697e6bc8513b72b2f5307e25fff9b",
"globalAliases": [],
"localAliases": [
{
"accessKeyId": "GK31c2f218a2e44f485b94239e",
"alias": "test"
}
]
}
]
```
#### GetBucketInfo `GET /v0/bucket?id=<bucket id>`
#### GetBucketInfo `GET /v0/bucket?globalAlias=<alias>`
Returns information about the requested storage bucket.
If `id` is set, the bucket is looked up using its exact identifier.
If `globalAlias` is set, the bucket is looked up using its global alias.
(both are fast)
Example response:
```json
{
"id": "afa8f0a22b40b1247ccd0affb869b0af5cff980924a20e4b5e0720a44deb8d39",
"globalAliases": [],
"websiteAccess": false,
"websiteConfig": null,
"keys": [
{
"accessKeyId": "GK31c2f218a2e44f485b94239e",
"name": "Imported key",
"permissions": {
"read": true,
"write": true,
"owner": true
},
"bucketLocalAliases": [
"debug"
]
}
],
"objects": 14827,
"bytes": 13189855625,
"unfinshedUploads": 0,
"quotas": {
"maxSize": null,
"maxObjects": null
}
}
```
#### CreateBucket `POST /v0/bucket`
Creates a new storage bucket.
Request body format:
```json
{
"globalAlias": "NameOfMyBucket"
}
```
OR
```json
{
"localAlias": {
"accessKeyId": "GK31c2f218a2e44f485b94239e",
"alias": "NameOfMyBucket",
"allow": {
"read": true,
"write": true,
"owner": false
}
}
}
```
OR
```json
{}
```
Creates a new bucket, either with a global alias, a local one,
or no alias at all.
Technically, you can also specify both `globalAlias` and `localAlias` and that would create
two aliases, but I don't see why you would want to do that.
#### DeleteBucket `DELETE /v0/bucket?id=<bucket id>`
Deletes a storage bucket. A bucket cannot be deleted if it is not empty.
Warning: this will delete all aliases associated with the bucket!
#### UpdateBucket `PUT /v0/bucket?id=<bucket id>`
Updates configuration of the given bucket.
Request body format:
```json
{
"websiteAccess": {
"enabled": true,
"indexDocument": "index.html",
"errorDocument": "404.html"
},
"quotas": {
"maxSize": 19029801,
"maxObjects": null,
}
}
```
All fields (`websiteAccess` and `quotas`) are optionnal.
If they are present, the corresponding modifications are applied to the bucket, otherwise nothing is changed.
In `websiteAccess`: if `enabled` is `true`, `indexDocument` must be specified.
The field `errorDocument` is optional, if no error document is set a generic
error message is displayed when errors happen. Conversely, if `enabled` is
`false`, neither `indexDocument` nor `errorDocument` must be specified.
In `quotas`: new values of `maxSize` and `maxObjects` must both be specified, or set to `null`
to remove the quotas. An absent value will be considered the same as a `null`. It is not possible
to change only one of the two quotas.
### Operations on permissions for keys on buckets
#### BucketAllowKey `POST /v0/bucket/allow`
Allows a key to do read/write/owner operations on a bucket.
Request body format:
```json
{
"bucketId": "e6a14cd6a27f48684579ec6b381c078ab11697e6bc8513b72b2f5307e25fff9b",
"accessKeyId": "GK31c2f218a2e44f485b94239e",
"permissions": {
"read": true,
"write": true,
"owner": true
},
}
```
Flags in `permissions` which have the value `true` will be activated.
Other flags will remain unchanged.
#### BucketDenyKey `POST /v0/bucket/deny`
Denies a key from doing read/write/owner operations on a bucket.
Request body format:
```json
{
"bucketId": "e6a14cd6a27f48684579ec6b381c078ab11697e6bc8513b72b2f5307e25fff9b",
"accessKeyId": "GK31c2f218a2e44f485b94239e",
"permissions": {
"read": false,
"write": false,
"owner": true
},
}
```
Flags in `permissions` which have the value `true` will be deactivated.
Other flags will remain unchanged.
### Operations on bucket aliases
#### GlobalAliasBucket `PUT /v0/bucket/alias/global?id=<bucket id>&alias=<global alias>`
Empty body. Creates a global alias for a bucket.
#### GlobalUnaliasBucket `DELETE /v0/bucket/alias/global?id=<bucket id>&alias=<global alias>`
Removes a global alias for a bucket.
#### LocalAliasBucket `PUT /v0/bucket/alias/local?id=<bucket id>&accessKeyId=<access key ID>&alias=<local alias>`
Empty body. Creates a local alias for a bucket in the namespace of a specific access key.
#### LocalUnaliasBucket `DELETE /v0/bucket/alias/local?id=<bucket id>&accessKeyId<access key ID>&alias=<local alias>`
Removes a local alias for a bucket in the namespace of a specific access key.
@@ -0,0 +1,10 @@
*.aux
*.bbl
*.blg
*.log
*.nav
*.out
*.snm
*.synctex.gz
*.toc
*.dvi
@@ -0,0 +1,8 @@
all:
pdflatex présentation.tex
clean:
rm -f *.aux *.bbl *.blg *.log *.nav *.out *.snm *.synctex.gz *.toc *.dvi présentation.pdf
clean_sauf_pdf:
rm -f *.aux *.bbl *.blg *.log *.nav *.out *.snm *.synctex.gz *.toc *.dvi
Binary file not shown.

After

Width:  |  Height:  |  Size: 61 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 196 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 105 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 2.5 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 73 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 199 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 41 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 52 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 174 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 126 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 20 KiB

@@ -0,0 +1,340 @@
\documentclass[11pt, aspectratio=1610]{beamer}
\usetheme{Warsaw}
\usepackage[utf8]{inputenc}
\usepackage[french]{babel}
\usepackage{amsmath}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{tikz}
\usepackage{graphicx}
\usepackage{xcolor}
\usepackage{setspace}
\usepackage{todonotes}
\presetkeys{todonotes}{inline}{}
\renewcommand{\baselinestretch}{1.25}
\definecolor{orange_garage}{RGB}{255,147,41}
\definecolor{gris_garage}{RGB}{78,78,78}
\author[Association Deuxfleurs]{~\linebreak Vincent Giraud}
\title[De l'auto-hébergement à l'entre-hébergement avec Garage]{De l'auto-hébergement à l'entre-hébergement :\\Garage, pour conserver ses données ensemble}
%\setbeamercovered{transparent}
%\setbeamertemplate{navigation symbols}{}
\date{Capitole du Libre 2022\linebreak
\scriptsize Samedi 19 novembre 2022\linebreak
}
\setbeamercolor{palette primary}{fg=gris_garage,bg=orange_garage}
\setbeamercolor{palette secondary}{fg=gris_garage,bg=gris_garage}
\setbeamercolor{palette tiertary}{fg=white,bg=gris_garage}
\setbeamercolor{palette quaternary}{fg=white,bg=gris_garage}
\setbeamercolor{navigation symbols}{fg=black, bg=white}
\setbeamercolor{navigation symbols dimmed}{fg=darkgray, bg=white}
\setbeamercolor{itemize item}{fg=gris_garage}
\setbeamertemplate{itemize item}[circle]
\addtobeamertemplate{navigation symbols}{}{%
\usebeamerfont{footline}%
\usebeamercolor[fg]{footline}%
\hspace{1em}%
\insertframenumber/\inserttotalframenumber
}
\setbeamertemplate{headline}
{%
\leavevmode%
\begin{beamercolorbox}[wd=.5\paperwidth,ht=2.5ex,dp=1.125ex]{section in head/foot}%
\hbox to .5\paperwidth{\hfil\insertsectionhead\hfil}
\end{beamercolorbox}%
\begin{beamercolorbox}[wd=.5\paperwidth,ht=2.5ex,dp=1.125ex]{subsection in head/foot}%
\hbox to .5\paperwidth{\hfil\insertsubsectionhead\hfil}
\end{beamercolorbox}%
}
\addtobeamertemplate{footnote}{}{\vspace{2ex}}
\begin{document}
\begin{frame}
\titlepage
\end{frame}
\section{Introduction}
\subsection{Présentation}
\begin{frame}
\begin{columns}
\column{0.5 \linewidth}
\begin{center}
\includegraphics[width=3.5cm]{deuxfleurs-logo.png}\linebreak
\texttt{https://deuxfleurs.fr}
\end{center}
\column{0.4 \linewidth}
\begin{center}
Deuxfleurs est une association militant en faveur d'un internet plus convivial, avec une organisation et des rapports de force repensés.\linebreak
Nous faisons partie du CHATONS\footnote[frame]{Collectif des Hébergeurs Alternatifs, Transparents, Ouverts, Neutres et Solidaires} depuis avril 2022.
\includegraphics[width=2cm]{logo_chatons.png}
\end{center}
\end{columns}
\end{frame}
\subsection{Héberger à la maison}
\begin{frame}
\begin{columns}
\begin{column}{0.5 \linewidth}
\begin{center}
Pour échapper au contrôle et au giron des opérateurs de clouds, héberger ses données à la maison présente de nombreux avantages...
\end{center}
\vspace{0.5cm}
\begin{itemize}[<+(1)->]
\item On récupère la souveraineté sur ses données
\item On gagne en vie privée
\item On gagne en libertés
\item On est responsabilisé face à ses besoins
\end{itemize}
\end{column}
\vrule{}
\begin{column}{0.5 \linewidth}
\begin{center}
\onslide<6->{... mais aussi bien des contraintes...}
\end{center}
\vspace{0.5cm}
\begin{itemize}[<+(2)->]
\item On repose sur une connexion internet pour particulier
\item Un certain savoir-faire et moultes compétences sont requis
\item Assurer la résilience de ses services est difficile
\item Bien sauvegarder ses données, et ceci au-delà de son site géographique, n'est pas évident
\end{itemize}
\end{column}
\end{columns}
\end{frame}
\subsection{Sauvegarder pour se parer à tout imprévu}
\begin{frame}
\begin{center}
Sauvegarder pour se parer contre les pannes matérielles est une chose...
Sauvegarder pour se parer contre les cambriolages et les incendies en est une autre !\linebreak
\vspace{1cm}
\onslide<2->{Répartir géographiquement ses données devient alors nécessaire.}
\end{center}
\end{frame}
\section{Les solutions à explorer}
\subsection{L'entre-hébergement}
\begin{frame}
\begin{center}
On a vu récemment se développer au sein du CHATONS la notion d'entre-hébergement : en plus de renforcer l'intégrité des sauvegardes, on va améliorer la disponibilité pendant les coupures de liaison internet, de courant, ou pendant les déménagements d'administrateurs par exemple.\linebreak
\vspace{1cm}
\onslide<2->
{
Dans le cadre du collectif, il s'agit de partager ses volumes de données entre hébergeurs.\linebreak
Pour assurer la confidentialité, on peut chiffrer les données au niveau applicatif.
}
\end{center}
\end{frame}
\subsection{S3 contre les systèmes de fichiers}
\begin{frame}
\begin{center}
Dans le cadre de l'administration de services en ligne, les systèmes de fichiers recèlent certaines difficultés.\linebreak
\vspace{1cm}
Le standard S3 apporte des facilités; on réduit le stockage à un paradigme de clé-valeur basé essentiellement sur deux opérations seulement: lire ou écrire une clé.
\end{center}
\end{frame}
\section{Garage}
\subsection{Présentation}
\begin{frame}
\begin{columns}
\column{0.5 \linewidth}
\begin{center}
Garage essaye de répondre à l'ensemble de ces besoins.\linebreak
\vspace{0.5cm}
Il s'agit d'un logiciel libre permettant de distribuer un service S3 sur diverses machines éloignées.
\end{center}
\column{0.5 \linewidth}
\begin{center}
\includegraphics[width=4cm]{garage-logo.png}\linebreak
\texttt{https://garagehq.deuxfleurs.fr/}
\end{center}
\end{columns}
\end{frame}
\subsection{Gestion des zones}
\begin{frame}
\begin{center}
Garage va prendre en compte les zones géographiques au moment de répliquer les données.\linebreak
\vspace{1cm}
\includegraphics[width=13.25cm]{zones.png}
\end{center}
\end{frame}
\subsection{Comment ça marche ?}
\begin{frame}
\begin{columns}
\column{0.5 \linewidth}
\input{schéma europe}
\column{0.5 \linewidth}
\begin{center}
Chaque objet est dupliqué sur plusieurs zones différentes.\linebreak
\onslide<5->{Lorsqu'un nouvel hébergeur rejoint le réseau, la charge se voit équilibrée.}\linebreak
\onslide<12->{Si une zone devient indisponible, les autres continuent d'assurer le service.}\linebreak
\end{center}
\end{columns}
\end{frame}
\subsection{Financement}
\begin{frame}
\begin{center}
Dans le cadre du programme \textit{Horizon 2021} de l'Union Européenne, nous avons reçu une subvention de la part de l'initiative NGI Pointer\footnote[frame]{Next Generation Internet Program for Open Internet Renovation}.\linebreak
\includegraphics[width=3cm]{drapeau_européen.png}\hspace{1cm}
\includegraphics[width=3cm]{NGI.png}\linebreak
Nous avons ainsi pu financer le développement de Garage pendant 1 an.
\end{center}
\end{frame}
\subsection{Licence}
\begin{frame}
\begin{center}
De par nos valeurs, nous avons attribué la licence AGPL version 3 à Garage, notamment afin qu'il reste parmi les biens communs.\linebreak
\vspace{0.5cm}
\includegraphics[width=5cm]{agpl-v3-logo.png}\linebreak
\end{center}
\end{frame}
\subsection{Langage utilisé}
\begin{frame}
\begin{center}
Nous avons décidé d'écrire Garage à l'aide du langage Rust, afin d'obtenir une compilation vers des binaires natifs et efficaces.\linebreak
\includegraphics[width=3.5cm]{rust-logo.png}\linebreak
Ce choix permet également de bénéficier des avantages reconnus de Rust en termes de sécurité.
\end{center}
\end{frame}
\subsection{Matériel requis}
\begin{frame}
\begin{center}
Garage peut ainsi être performant sur des machines limitées. Les prérequis sont minimes : n'importe quelle machine avec un processeur qui a moins d'une décennie, 1~gigaoctet de mémoire vive, et 16~gigaoctets de stockage suffit.\linebreak
\vspace{1cm}
Cet aspect est déterminant : il permet en effet d'héberger sur du matériel acheté d'occasion, pour réduire l'impact écologique de nos infrastructures.
\end{center}
\end{frame}
\subsection{Performances}
\begin{frame}
\begin{center}
\includegraphics[width=13.25cm]{rpc-amplification.png}
\end{center}
\end{frame}
\begin{frame}
\begin{center}
\includegraphics[width=11cm]{rpc-complexity.png}
\end{center}
\end{frame}
\subsection{Services}
\begin{frame}
\begin{center}
Puisqu'il suit le standard S3, beaucoup de services populaires sont par conséquence compatibles avec Garage :\linebreak
\begin{columns}
\column{0.2 \linewidth}
\begin{center}
\includegraphics[width=2.5cm]{nextcloud-logo.png}
\end{center}
\column{0.2 \linewidth}
\begin{center}
\includegraphics[width=2.5cm]{peertube-logo.png}
\end{center}
\column{0.2 \linewidth}
\begin{center}
\includegraphics[width=2.5cm]{matrix-logo.png}
\end{center}
\column{0.2 \linewidth}
\begin{center}
\includegraphics[width=2.5cm]{mastodon-logo.png}
\end{center}
\end{columns}
~\linebreak
Et comme souvent avec S3, on peut assimiler un bucket à un site, et utiliser le serveur pour héberger des sites web statiques.
\end{center}
\end{frame}
\section{Intégration chez Deuxfleurs}
\subsection{Matériel}
\begin{frame}
\begin{center}
\includegraphics[width=13cm]{neptune.jpg}\linebreak
En pratique, nos serveurs ne sont effectivement que des machines achetées d'occasion (très souvent des anciens ordinateurs destinés à la bureautique en entreprise).
\end{center}
\end{frame}
\subsection{Environnement logiciel}
\begin{frame}
\begin{center}
Pour faciliter la reproduction d'un environnement connu, NixOS est installé sur nos machines.\linebreak
\vspace{1cm}
Pour saccommoder des réseaux qu'on trouve derrière des routeurs pour particuliers, on s'aide de notre logiciel Diplonat\footnote[frame]{\texttt{https://git.deuxfleurs.fr/Deuxfleurs/diplonat}}.
\end{center}
\end{frame}
\section{Au-delà...}
\subsection{... de Deuxfleurs}
\begin{frame}
\begin{center}
\includegraphics[width=10cm]{tedomum.png}
\end{center}
\end{frame}
\subsection{... de Garage}
\begin{frame}
\begin{center}
Nous avons récemment lancé le développement d'Aérogramme\footnote[frame]{\texttt{https://git.deuxfleurs.fr/Deuxfleurs/aerogramme}}.\linebreak
\vspace{1cm}
Il s'agit d'un serveur de stockage de courriels chiffrés.\linebreak
\vspace{1cm}
Il est conçu pour pouvoir travailler avec Garage.
\end{center}
\end{frame}
\section{Fin}
\subsection{Contacts}
\begin{frame}
\begin{center}
\begin{tikzpicture}
\node (ronce) {\includegraphics[width=0.95\textwidth]{ronce.jpg}};
\node[white] at (-5.1,3.6) {Intéressé(e) ?};
\node[white, align=center] at (4.2,-2.6) {Contactez-nous !\\\texttt{coucou@deuxfleurs.fr}\\\texttt{\#forum:deuxfleurs.fr}};
\end{tikzpicture}
\end{center}
\end{frame}
\end{document}
Binary file not shown.

After

Width:  |  Height:  |  Size: 1.4 MiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 124 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 194 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 6.5 KiB

@@ -0,0 +1,52 @@
\begin{tikzpicture}
\node (carte) {\includegraphics[width=\textwidth]{carte-Europe.pdf}};
% \personnage{position X}{position Y}{facteur d'échelle}
\newcommand{\personnage}[4]
{
\fill[#4] ({#1-(0.4 * #3)},{#2-(0.9 * #3)}) .. controls ({#1-(0.4 * #3)},#2) and ({#1+(0.4 * #3)},#2) .. ({#1+(0.4 * #3)},{#2-(0.9 * #3)}) -- ({#1-(0.4 * #3)},{#2-(0.9 * #3)});
\fill[#4] (#1,#2) circle ({0.25 * #3});
}
\onslide<1-11>{\personnage{-2.25}{-0.75}{0.75}{green}}
\onslide<1-11>{\draw (-1.9,-1.6) rectangle ++(1,1.2);}
\onslide<2-11>{\draw[fill=green] (-1.8,-1.525) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 1};}
\onslide<4-5>{\draw[fill=red] (-1.8,-1.15) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 3};}
\onslide<7-11>{\draw[fill=yellow] (-1.8,-1.15) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 4};}
\onslide<9-11>{\draw[fill=red] (-1.8,-0.775) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 6};}
\onslide<3-11>{\draw[fill=blue] (-1.35,-1.525) rectangle ++(0.35,0.3) node[pos=0.5, white] {\tiny 2};}
\onslide<8-11>{\draw[fill=blue] (-1.35,-1.15) rectangle ++(0.35,0.3) node[pos=0.5, white] {\tiny 5};}
\onslide<11-11>{\draw[fill=yellow] (-1.35,-0.775) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 8};}
\personnage{1.65}{1.5}{0.75}{blue}
\draw (0.3,0.7) rectangle ++(1,1.2);
\onslide<2->{\draw[fill=green] (0.4,0.775) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 1};}
\onslide<4->{\draw[fill=red] (0.4,1.15) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 3};}
\onslide<10->{\draw[fill=green] (0.4,1.525) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 7};}
\onslide<3->{\draw[fill=blue] (0.85,0.775) rectangle ++(0.35,0.3) node[pos=0.5, white] {\tiny 2};}
\onslide<9->{\draw[fill=red] (0.85,1.15) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 6};}
\onslide<11->{\draw[fill=yellow] (0.85,1.525) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 8};}
\personnage{1.85}{-2.3}{0.75}{red}
\draw (0.5,-3.15) rectangle ++(1,1.2);
\onslide<2->{\draw[fill=green] (0.6,-3.075) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 1};}
\onslide<4-5>{\draw[fill=red] (0.6,-2.7) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 3};}
\onslide<7->{\draw[fill=yellow] (0.6,-2.7) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 4};}
\onslide<9->{\draw[fill=red] (0.6,-2.325) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 6};}
\onslide<3-5>{\draw[fill=blue] (1.05,-3.075) rectangle ++(0.35,0.3) node[pos=0.5, white] {\tiny 2};}
\onslide<6->{\draw[fill=red] (1.05,-3.075) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 3};}
\onslide<8->{\draw[fill=blue] (1.05,-2.7) rectangle ++(0.35,0.3) node[pos=0.5, white] {\tiny 5};}
\onslide<10->{\draw[fill=green] (1.05,-2.325) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 7};}
\onslide<5->{\personnage{1.05}{-0.15}{0.75}{yellow}}
\onslide<5->{\draw (-0.35,-1) rectangle ++(1,1.2);}
\onslide<6->{\draw[fill=blue] (-0.25,-0.925) rectangle ++(0.35,0.3) node[pos=0.5, white] {\tiny 2};}
\onslide<7->{\draw[fill=yellow] (-0.25,-0.55) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 4};}
\onslide<10->{\draw[fill=green] (-0.25,-0.175) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 7};}
\onslide<6->{\draw[fill=red] (0.2,-0.925) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 3};}
\onslide<8->{\draw[fill=blue] (0.2,-0.55) rectangle ++(0.35,0.3) node[pos=0.5,white] {\tiny 5};}
\onslide<11->{\draw[fill=yellow] (0.2,-0.175) rectangle ++(0.35,0.3) node[pos=0.5] {\tiny 8};}
\onslide<12->{\draw[line width=0.25cm] (-2.15,-0.5) -- ++(1,-1);}
\onslide<12->{\draw[line width=0.25cm] (-2.15,-1.5) -- ++(1,1);}
\end{tikzpicture}
Binary file not shown.

After

Width:  |  Height:  |  Size: 236 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 97 KiB

Generated
+108
View File
@@ -0,0 +1,108 @@
{
"nodes": {
"cargo2nix": {
"inputs": {
"flake-compat": "flake-compat",
"flake-utils": "flake-utils",
"nixpkgs": [
"nixpkgs"
],
"rust-overlay": "rust-overlay"
},
"locked": {
"lastModified": 1673262828,
"narHash": "sha256-pDqno5/2ghQDt4LjVt5ZUMV9pTSA5rGGdz6Skf2rBwc=",
"owner": "Alexis211",
"repo": "cargo2nix",
"rev": "505caa32110d42ee03bd68b47031142eff9c827b",
"type": "github"
},
"original": {
"owner": "Alexis211",
"repo": "cargo2nix",
"rev": "505caa32110d42ee03bd68b47031142eff9c827b",
"type": "github"
}
},
"flake-compat": {
"flake": false,
"locked": {
"lastModified": 1650374568,
"narHash": "sha256-Z+s0J8/r907g149rllvwhb4pKi8Wam5ij0st8PwAh+E=",
"owner": "edolstra",
"repo": "flake-compat",
"rev": "b4a34015c698c7793d592d66adbab377907a2be8",
"type": "github"
},
"original": {
"owner": "edolstra",
"repo": "flake-compat",
"type": "github"
}
},
"flake-utils": {
"locked": {
"lastModified": 1659877975,
"narHash": "sha256-zllb8aq3YO3h8B/U0/J1WBgAL8EX5yWf5pMj3G0NAmc=",
"owner": "numtide",
"repo": "flake-utils",
"rev": "c0e246b9b83f637f4681389ecabcb2681b4f3af0",
"type": "github"
},
"original": {
"owner": "numtide",
"repo": "flake-utils",
"type": "github"
}
},
"nixpkgs": {
"locked": {
"lastModified": 1673261889,
"narHash": "sha256-7trMsi0z7EfYNC/Nc5EtulvChBHQAo376XRICyWr89Q=",
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "baed728abe983508cabc99d05cccc164fe748744",
"type": "github"
},
"original": {
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "baed728abe983508cabc99d05cccc164fe748744",
"type": "github"
}
},
"root": {
"inputs": {
"cargo2nix": "cargo2nix",
"nixpkgs": "nixpkgs"
}
},
"rust-overlay": {
"inputs": {
"flake-utils": [
"cargo2nix",
"flake-utils"
],
"nixpkgs": [
"cargo2nix",
"nixpkgs"
]
},
"locked": {
"lastModified": 1664247556,
"narHash": "sha256-J4vazHU3609ekn7dr+3wfqPo5WGlZVAgV7jfux352L0=",
"owner": "oxalica",
"repo": "rust-overlay",
"rev": "524db9c9ea7bc7743bb74cdd45b6d46ea3fcc2ab",
"type": "github"
},
"original": {
"owner": "oxalica",
"repo": "rust-overlay",
"type": "github"
}
}
},
"root": "root",
"version": 7
}
+28
View File
@@ -0,0 +1,28 @@
{
description = "Garage, an S3-compatible distributed object store for self-hosted deployments";
inputs.nixpkgs.url = "github:NixOS/nixpkgs/baed728abe983508cabc99d05cccc164fe748744";
inputs.cargo2nix = {
# As of 2022-10-18: two small patches over unstable branch, one for clippy and one to fix feature detection
url = "github:Alexis211/cargo2nix/505caa32110d42ee03bd68b47031142eff9c827b";
inputs.nixpkgs.follows = "nixpkgs";
};
outputs = { self, nixpkgs, cargo2nix }: let
git_version = self.lastModifiedDate;
compile = import ./nix/compile.nix;
forAllSystems = nixpkgs.lib.genAttrs nixpkgs.lib.systems.flakeExposed;
in
{
packages = forAllSystems (system: {
default = (compile {
inherit system git_version;
pkgsSrc = nixpkgs;
cargo2nixOverlay = cargo2nix.overlays.default;
release = true;
}).workspace.garage {
compileMode = "build";
};
});
};
}
+4 -4
View File
@@ -3,15 +3,15 @@ rec {
* Fixed dependencies
*/
pkgsSrc = fetchTarball {
# As of 2022-10-13
url = "https://github.com/NixOS/nixpkgs/archive/a3073c49bc0163fea6a121c276f526837672b555.zip";
sha256 = "1bz632psfbpmicyzjb8b4265y50shylccvfm6ry6mgnv5hvz324s";
# As of 2023-01-09
url = "https://github.com/NixOS/nixpkgs/archive/baed728abe983508cabc99d05cccc164fe748744.zip";
sha256 = "1m7kmcjhnj3lx7xqs0nh262c4nxs5n8p7k9g6kc4gv1k5nrcrnpf";
};
cargo2nixSrc = fetchGit {
# As of 2022-10-18: two small patches over unstable branch, one for clippy and one to fix feature detection
url = "https://github.com/Alexis211/cargo2nix";
ref = "custom_unstable";
rev = "a7a61179b66054904ef6a195d8da736eaaa06c36";
rev = "505caa32110d42ee03bd68b47031142eff9c827b";
};
/*
+21 -14
View File
@@ -1,25 +1,32 @@
{
system ? builtins.currentSystem,
target,
system,
target ? null,
pkgsSrc,
cargo2nixOverlay,
compiler ? "rustc",
release ? false,
git_version ? null,
features ? null,
}:
with import ./common.nix;
let
log = v: builtins.trace v v;
pkgs = import pkgsSrc {
inherit system;
crossSystem = {
config = target;
isStatic = true;
};
overlays = [ cargo2nixOverlay ];
};
pkgs =
if target != null then
import pkgsSrc {
inherit system;
crossSystem = {
config = target;
isStatic = true;
};
overlays = [ cargo2nixOverlay ];
}
else
import pkgsSrc {
inherit system;
overlays = [ cargo2nixOverlay ];
};
/*
Cargo2nix is built for rustOverlay which installs Rust from Mozilla releases.
@@ -34,8 +41,8 @@ let
NixOS ships them in separate ones. We reunite them with symlinkJoin.
*/
toolchainOptions =
if target == "x86_64-unknown-linux-musl" || target == "aarch64-unknown-linux-musl" then {
rustVersion = "1.63.0";
if target == null || target == "x86_64-unknown-linux-musl" || target == "aarch64-unknown-linux-musl" then {
rustVersion = "1.65.0";
extraRustComponents = [ "clippy" ];
} else {
rustToolchain = pkgs.symlinkJoin {
+1 -1
View File
@@ -7,7 +7,7 @@ pkgs.buildGoModule rec {
owner = "GoogleContainerTools";
repo = "kaniko";
rev = "v${version}";
sha256 = "1fnclr556avxay6pvgw5ya3xbxfnf2gv4njq2hr4fd6fcjyslq5h";
sha256 = "TXgzO/NfLXVo5a7yyO3XYSk+9H1CwMF+vwbRx3kchQ8=";
};
vendorSha256 = null;
+1 -1
View File
@@ -15,7 +15,7 @@ type: application
# This is the chart version. This version number should be incremented each time you make changes
# to the chart and its templates, including the app version.
# Versions are expected to follow Semantic Versioning (https://semver.org/)
version: 0.1.0
version: 0.2.0
# This is the version number of the application being deployed. This version number should be
# incremented each time you make changes to the application. Versions are not expected to
@@ -18,6 +18,9 @@ metadata:
name: {{ $fullName }}-s3-api
labels:
{{- include "garage.labels" . | nindent 4 }}
{{- with .Values.ingress.s3.api.labels }}
{{- toYaml . | nindent 4 }}
{{- end }}
{{- with .Values.ingress.s3.api.annotations }}
annotations:
{{- toYaml . | nindent 4 }}
@@ -80,6 +83,9 @@ metadata:
name: {{ $fullName }}-s3-web
labels:
{{- include "garage.labels" . | nindent 4 }}
{{- with .Values.ingress.s3.web.labels }}
{{- toYaml . | nindent 4 }}
{{- end }}
{{- with .Values.ingress.s3.web.annotations }}
annotations:
{{- toYaml . | nindent 4 }}
+10 -5
View File
@@ -85,14 +85,15 @@ service:
ingress:
s3:
api:
enabled: true
enabled: false
# Rely either on the className or the annotation below but not both
# replace "nginx" by an Ingress controller
# you can find examples here https://kubernetes.io/docs/concepts/services-networking/ingress-controllers
className: "nginx"
annotations:
# className: "nginx"
annotations: {}
# kubernetes.io/ingress.class: "nginx"
# kubernetes.io/tls-acme: "true"
labels: {}
hosts:
- host: "s3.garage.tld" # garage S3 API endpoint
paths:
@@ -107,11 +108,15 @@ ingress:
# hosts:
# - kubernetes.docker.internal
web:
enabled: true
className: "nginx"
enabled: false
# Rely either on the className or the annotation below but not both
# replace "nginx" by an Ingress controller
# you can find examples here https://kubernetes.io/docs/concepts/services-networking/ingress-controllers
# className: "nginx"
annotations: {}
# kubernetes.io/ingress.class: nginx
# kubernetes.io/tls-acme: "true"
labels: {}
hosts:
- host: "*.web.garage.tld" # wildcard website access with bucket name prefix
paths:
+14 -2
View File
@@ -1,5 +1,5 @@
{
system ? builtins.currentSystem,
system ? builtins.currentSystem,
}:
with import ./nix/common.nix;
@@ -71,13 +71,25 @@ function refresh_cache {
for attr in clippy.amd64 test.amd64 pkgs.{amd64,i386,arm,arm64}.{debug,release}; do
echo "Updating cache for ''${attr}"
derivation=$(nix-instantiate --attr ''${attr})
nix copy \
nix copy -j8 \
--to 's3://nix?endpoint=garage.deuxfleurs.fr&region=garage&secret-key=/tmp/nix-signing-key.sec' \
$(nix-store -qR ''${derivation%\!bin})
done
rm /tmp/nix-signing-key.sec
}
function refresh_flake_cache {
pass show deuxfleurs/nix_priv_key > /tmp/nix-signing-key.sec
for attr in packages.x86_64-linux.default; do
echo "Updating cache for ''${attr}"
derivation=$(nix path-info --derivation ".#''${attr}")
nix copy -j8 \
--to 's3://nix?endpoint=garage.deuxfleurs.fr&region=garage&secret-key=/tmp/nix-signing-key.sec' \
$(nix-store -qR ''${derivation})
done
rm /tmp/nix-signing-key.sec
}
function to_s3 {
aws \
--endpoint-url https://garage.deuxfleurs.fr \
+6 -6
View File
@@ -1,6 +1,6 @@
[package]
name = "garage_api"
version = "0.8.0"
version = "0.8.1"
authors = ["Alex Auvolat <alex@adnab.me>"]
edition = "2018"
license = "AGPL-3.0"
@@ -14,11 +14,11 @@ path = "lib.rs"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
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" }
garage_model = { version = "0.8.1", path = "../model" }
garage_table = { version = "0.8.1", path = "../table" }
garage_block = { version = "0.8.1", path = "../block" }
garage_util = { version = "0.8.1", path = "../util" }
garage_rpc = { version = "0.8.1", path = "../rpc" }
async-trait = "0.1.7"
base64 = "0.13"
+29
View File
@@ -15,6 +15,7 @@ use opentelemetry_prometheus::PrometheusExporter;
use prometheus::{Encoder, TextEncoder};
use garage_model::garage::Garage;
use garage_rpc::system::ClusterHealthStatus;
use garage_util::error::Error as GarageError;
use crate::generic_server::*;
@@ -76,6 +77,31 @@ impl AdminApiServer {
.body(Body::empty())?)
}
fn handle_health(&self) -> Result<Response<Body>, Error> {
let health = self.garage.system.health();
let (status, status_str) = match health.status {
ClusterHealthStatus::Healthy => (StatusCode::OK, "Garage is fully operational"),
ClusterHealthStatus::Degraded => (
StatusCode::OK,
"Garage is operational but some storage nodes are unavailable",
),
ClusterHealthStatus::Unavailable => (
StatusCode::SERVICE_UNAVAILABLE,
"Quorum is not available for some/all partitions, reads and writes will fail",
),
};
let status_str = format!(
"{}\nConsult the full health check API endpoint at /v0/health for more details\n",
status_str
);
Ok(Response::builder()
.status(status)
.header(http::header::CONTENT_TYPE, "text/plain")
.body(Body::from(status_str))?)
}
fn handle_metrics(&self) -> Result<Response<Body>, Error> {
#[cfg(feature = "metrics")]
{
@@ -124,6 +150,7 @@ impl ApiHandler for AdminApiServer {
) -> Result<Response<Body>, Error> {
let expected_auth_header =
match endpoint.authorization_type() {
Authorization::None => None,
Authorization::MetricsToken => self.metrics_token.as_ref(),
Authorization::AdminToken => match &self.admin_token {
None => return Err(Error::forbidden(
@@ -147,8 +174,10 @@ impl ApiHandler for AdminApiServer {
match endpoint {
Endpoint::Options => self.handle_options(&req),
Endpoint::Health => self.handle_health(),
Endpoint::Metrics => self.handle_metrics(),
Endpoint::GetClusterStatus => handle_get_cluster_status(&self.garage).await,
Endpoint::GetClusterHealth => handle_get_cluster_health(&self.garage).await,
Endpoint::ConnectClusterNodes => handle_connect_cluster_nodes(&self.garage, req).await,
// Layout
Endpoint::GetClusterLayout => handle_get_cluster_layout(&self.garage).await,
+5
View File
@@ -43,6 +43,11 @@ pub async fn handle_get_cluster_status(garage: &Arc<Garage>) -> Result<Response<
Ok(json_ok_response(&res)?)
}
pub async fn handle_get_cluster_health(garage: &Arc<Garage>) -> Result<Response<Body>, Error> {
let health = garage.system.health();
Ok(json_ok_response(&health)?)
}
pub async fn handle_connect_cluster_nodes(
garage: &Arc<Garage>,
req: Request<Body>,
+15 -5
View File
@@ -6,6 +6,7 @@ use crate::admin::error::*;
use crate::router_macros::*;
pub enum Authorization {
None,
MetricsToken,
AdminToken,
}
@@ -16,8 +17,10 @@ router_match! {@func
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Endpoint {
Options,
Health,
Metrics,
GetClusterStatus,
GetClusterHealth,
ConnectClusterNodes,
// Layout
GetClusterLayout,
@@ -88,8 +91,10 @@ impl Endpoint {
let res = router_match!(@gen_path_parser (req.method(), path, query) [
OPTIONS _ => Options,
GET "/health" => Health,
GET "/metrics" => Metrics,
GET "/v0/status" => GetClusterStatus,
GET "/v0/health" => GetClusterHealth,
POST "/v0/connect" => ConnectClusterNodes,
// Layout endpoints
GET "/v0/layout" => GetClusterLayout,
@@ -130,6 +135,7 @@ impl Endpoint {
/// Get the kind of authorization which is required to perform the operation.
pub fn authorization_type(&self) -> Authorization {
match self {
Self::Health => Authorization::None,
Self::Metrics => Authorization::MetricsToken,
_ => Authorization::AdminToken,
}
@@ -137,9 +143,13 @@ impl Endpoint {
}
generateQueryParameters! {
"id" => id,
"search" => search,
"globalAlias" => global_alias,
"alias" => alias,
"accessKeyId" => access_key_id
keywords: [],
fields: [
"format" => format,
"id" => id,
"search" => search,
"globalAlias" => global_alias,
"alias" => alias,
"accessKeyId" => access_key_id
]
}
+1 -1
View File
@@ -211,7 +211,7 @@ pub async fn handle_poll_item(
let item = garage
.k2v
.rpc
.poll(
.poll_item(
bucket_id,
partition_key,
sort_key,
+20 -23
View File
@@ -96,7 +96,7 @@ impl Endpoint {
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),
(query.keyword.take().unwrap_or_default(), partition_key, query, None),
key: [
EMPTY if causality_token => PollItem (query::sort_key, query::causality_token, opt_parse::timeout),
EMPTY => ReadItem (query::sort_key),
@@ -111,7 +111,7 @@ impl Endpoint {
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),
(query.keyword.take().unwrap_or_default(), partition_key, query, None),
key: [
],
no_key: [
@@ -125,7 +125,7 @@ impl Endpoint {
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),
(query.keyword.take().unwrap_or_default(), partition_key, query, None),
key: [
EMPTY => HeadObject(opt_parse::part_number, query_opt::version_id),
],
@@ -140,7 +140,7 @@ impl Endpoint {
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),
(query.keyword.take().unwrap_or_default(), partition_key, query, None),
key: [
],
no_key: [
@@ -155,7 +155,7 @@ impl Endpoint {
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),
(query.keyword.take().unwrap_or_default(), partition_key, query, None),
key: [
EMPTY => InsertItem (query::sort_key),
@@ -169,7 +169,7 @@ impl Endpoint {
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),
(query.keyword.take().unwrap_or_default(), partition_key, query, None),
key: [
EMPTY => DeleteItem (query::sort_key),
],
@@ -232,21 +232,18 @@ impl Endpoint {
// 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";
keywords: [
"delete" => DELETE,
"search" => SEARCH
],
fields: [
"prefix" => prefix,
"start" => start,
"causality_token" => causality_token,
"end" => end,
"limit" => limit,
"reverse" => reverse,
"sort_key" => sort_key,
"timeout" => timeout
]
}
+74 -58
View File
@@ -4,10 +4,9 @@ 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 {
$(
$endpoint { .. } => true,
Endpoint::$endpoint { .. } => true,
)*
_ => false
}
@@ -15,37 +14,35 @@ macro_rules! router_match {
(@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($param),
Endpoint::$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, $param),
)*)?
},
)*
(m, p) => {
return Err(Error::bad_request(format!(
"Unknown API endpoint: {} {}",
m, p
)))
}
}
}
}};
(@gen_path_parser ($method:expr, $reqpath:expr, $query:expr)
[
$($meth:ident $path:pat $(if $required:ident)? => $api:ident $(($($conv:ident :: $param:ident),*))?,)*
]) => {{
{
match ($method, $reqpath) {
$(
(&Method::$meth, $path) if true $(&& $query.$required.is_some())? => Endpoint::$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 :: $param_nk:ident),*))?,)*]) => {{
@@ -60,11 +57,9 @@ macro_rules! router_match {
// ]
// }
// See in from_{method} for more detailed usage.
use Endpoint::*;
use keywords::*;
match ($keyword, !$key.is_empty()){
$(
($kw_k, true) if true $(&& $query.$required_k.is_some())? $(&& $header.contains_key($header_k))? => Ok($api_k {
(Keyword::$kw_k, true) if true $(&& $query.$required_k.is_some())? $(&& $header.contains_key($header_k))? => Ok(Endpoint::$api_k {
$key,
$($(
$param_k: router_match!(@@parse_param $query, $conv_k, $param_k),
@@ -72,7 +67,7 @@ macro_rules! router_match {
}),
)*
$(
($kw_nk, false) $(if $query.$required_nk.is_some())? $(if $header.contains($header_nk))? => Ok($api_nk {
(Keyword::$kw_nk, false) $(if $query.$required_nk.is_some())? $(if $header.contains($header_nk))? => Ok(Endpoint::$api_nk {
$($(
$param_nk: router_match!(@@parse_param $query, $conv_nk, $param_nk),
)*)?
@@ -84,7 +79,7 @@ macro_rules! router_match {
(@@parse_param $query:expr, query_opt, $param:ident) => {{
// extract optional query parameter
$query.$param.take().map(|param| param.into_owned())
$query.$param.take().map(|param| param.into_owned())
}};
(@@parse_param $query:expr, query, $param:ident) => {{
// extract mendatory query parameter
@@ -93,7 +88,7 @@ macro_rules! router_match {
(@@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
$query.$param
.take()
.map(|param| param.parse())
.transpose()
@@ -144,14 +139,40 @@ macro_rules! router_match {
/// 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),* ) => {
(
keywords: [ $($kw_param:expr => $kw_name: ident),* ],
fields: [ $($f_param:expr => $f_name:ident),* ]
) => {
#[derive(Debug)]
#[allow(non_camel_case_types)]
#[allow(clippy::upper_case_acronyms)]
enum Keyword {
EMPTY,
$( $kw_name, )*
}
impl std::fmt::Display for Keyword {
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
match self {
Keyword::EMPTY => write!(f, "``"),
$( Keyword::$kw_name => write!(f, "`{}`", $kw_param), )*
}
}
}
impl Default for Keyword {
fn default() -> Self {
Keyword::EMPTY
}
}
/// 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>>,
keyword: Option<Keyword>,
$(
$name: Option<Cow<'a, str>>,
$f_name: Option<Cow<'a, str>>,
)*
}
@@ -160,34 +181,29 @@ macro_rules! generateQueryParameters {
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() {
match k.as_ref() {
$(
$rest => if !v.is_empty() {
res.$name.replace(v).is_some()
} else {
false
$kw_param => if let Some(prev_kw) = res.keyword.replace(Keyword::$kw_name) {
return Err(Error::bad_request(format!(
"Multiple keywords: '{}' and '{}'", prev_kw, $kw_param
)));
},
)*
$(
$f_param => if !v.is_empty() {
if res.$f_name.replace(v).is_some() {
return Err(Error::bad_request(format!(
"Query parameter repeated: '{}'", k
)));
}
},
)*
_ => {
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 {
if !(k.starts_with("response-") || k.starts_with("X-Amz-")) {
debug!("Received an unknown query parameter: '{}'", k);
false
}
}
};
if repeated {
return Err(Error::bad_request(format!(
"Query parameter repeated: '{}'",
k
)));
}
}
Ok(res)
}
@@ -198,8 +214,8 @@ macro_rules! generateQueryParameters {
if self.keyword.is_some() {
Some("Keyword not used")
} $(
else if self.$name.is_some() {
Some(concat!("'", $rest, "'"))
else if self.$f_name.is_some() {
Some(concat!("'", $f_param, "'"))
}
)* else {
None
+9
View File
@@ -161,6 +161,15 @@ pub async fn handle_create_bucket(
return Err(CommonError::BucketAlreadyExists.into());
}
} else {
// Check user is allowed to create bucket
if !key_params.allow_create_bucket.get() {
return Err(CommonError::Forbidden(format!(
"Access key {} is not allowed to create buckets",
api_key.key_id
))
.into());
}
// Create the bucket!
if !is_valid_bucket_name(&bucket_name) {
return Err(Error::bad_request(format!(
+59 -34
View File
@@ -119,6 +119,17 @@ pub(crate) async fn save_stream<S: Stream<Item = Result<Bytes, Error>> + Unpin>(
return Ok((version_uuid, data_md5sum_hex));
}
// The following consists in many steps that can each fail.
// Keep track that some cleanup will be needed if things fail
// before everything is finished (cleanup is done using the Drop trait).
let mut interrupted_cleanup = InterruptedCleanup(Some((
garage.clone(),
bucket.id,
key.into(),
version_uuid,
version_timestamp,
)));
// Write version identifier in object table so that we have a trace
// that we are uploading something
let mut object_version = ObjectVersion {
@@ -139,44 +150,27 @@ pub(crate) async fn save_stream<S: Stream<Item = Result<Bytes, Error>> + Unpin>(
// Transfer data and verify checksum
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?;
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,
)?;
ensure_checksum_matches(
data_md5sum.as_slice(),
data_sha256sum,
content_md5.as_deref(),
content_sha256,
)?;
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 {
Ok(rv) => rv,
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()]);
garage.object_table.insert(&object).await?;
return Err(e);
}
};
check_quotas(&garage, bucket, key, total_size).await?;
// Save final object state, marked as Complete
let md5sum_hex = hex::encode(md5sum_arr);
let md5sum_hex = hex::encode(data_md5sum);
object_version.state = ObjectVersionState::Complete(ObjectVersionData::FirstBlock(
ObjectVersionMeta {
headers,
@@ -188,6 +182,10 @@ pub(crate) async fn save_stream<S: Stream<Item = Result<Bytes, Error>> + Unpin>(
let object = Object::new(bucket.id, key.into(), vec![object_version]);
garage.object_table.insert(&object).await?;
// We were not interrupted, everything went fine.
// We won't have to clean up on drop.
interrupted_cleanup.cancel();
Ok((version_uuid, md5sum_hex))
}
@@ -426,6 +424,33 @@ pub fn put_response(version_uuid: Uuid, md5sum_hex: String) -> Response<Body> {
.unwrap()
}
struct InterruptedCleanup(Option<(Arc<Garage>, Uuid, String, Uuid, u64)>);
impl InterruptedCleanup {
fn cancel(&mut self) {
drop(self.0.take());
}
}
impl Drop for InterruptedCleanup {
fn drop(&mut self) {
if let Some((garage, bucket_id, key, version_uuid, version_ts)) = self.0.take() {
tokio::spawn(async move {
let object_version = ObjectVersion {
uuid: version_uuid,
timestamp: version_ts,
state: ObjectVersionState::Aborted,
};
let object = Object::new(bucket_id, key, vec![object_version]);
if let Err(e) = garage.object_table.insert(&object).await {
warn!("Cannot cleanup after aborted PutObject: {}", e);
}
});
}
}
}
// ----
pub async fn handle_create_multipart_upload(
garage: Arc<Garage>,
req: &Request<Body>,
+59 -62
View File
@@ -355,7 +355,7 @@ impl Endpoint {
fn from_get(key: String, query: &mut QueryParameters<'_>) -> Result<Self, Error> {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), key, query, None),
(query.keyword.take().unwrap_or_default(), key, query, None),
key: [
EMPTY if upload_id => ListParts (query::upload_id, opt_parse::max_parts, opt_parse::part_number_marker),
EMPTY => GetObject (query_opt::version_id, opt_parse::part_number),
@@ -412,7 +412,7 @@ impl Endpoint {
fn from_head(key: String, query: &mut QueryParameters<'_>) -> Result<Self, Error> {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), key, query, None),
(query.keyword.take().unwrap_or_default(), key, query, None),
key: [
EMPTY => HeadObject(opt_parse::part_number, query_opt::version_id),
],
@@ -426,7 +426,7 @@ impl Endpoint {
fn from_post(key: String, query: &mut QueryParameters<'_>) -> Result<Self, Error> {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), key, query, None),
(query.keyword.take().unwrap_or_default(), key, query, None),
key: [
EMPTY if upload_id => CompleteMultipartUpload (query::upload_id),
RESTORE => RestoreObject (query_opt::version_id),
@@ -448,7 +448,7 @@ impl Endpoint {
) -> Result<Self, Error> {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), key, query, headers),
(query.keyword.take().unwrap_or_default(), key, query, headers),
key: [
EMPTY if part_number header "x-amz-copy-source" => UploadPartCopy (parse::part_number, query::upload_id),
EMPTY header "x-amz-copy-source" => CopyObject,
@@ -490,7 +490,7 @@ impl Endpoint {
fn from_delete(key: String, query: &mut QueryParameters<'_>) -> Result<Self, Error> {
router_match! {
@gen_parser
(query.keyword.take().unwrap_or_default().as_ref(), key, query, None),
(query.keyword.take().unwrap_or_default(), key, query, None),
key: [
EMPTY if upload_id => AbortMultipartUpload (query::upload_id),
EMPTY => DeleteObject (query_opt::version_id),
@@ -624,63 +624,60 @@ impl Endpoint {
// parameter name => struct field
generateQueryParameters! {
"continuation-token" => continuation_token,
"delimiter" => delimiter,
"encoding-type" => encoding_type,
"fetch-owner" => fetch_owner,
"id" => id,
"key-marker" => key_marker,
"list-type" => list_type,
"marker" => marker,
"max-keys" => max_keys,
"max-parts" => max_parts,
"max-uploads" => max_uploads,
"partNumber" => part_number,
"part-number-marker" => part_number_marker,
"prefix" => prefix,
"select-type" => select_type,
"start-after" => start_after,
"uploadId" => upload_id,
"upload-id-marker" => upload_id_marker,
"versionId" => version_id,
"version-id-marker" => version_id_marker
}
mod keywords {
//! This module contain all query parameters with no associated value S3 uses to differentiate
//! endpoints.
pub const EMPTY: &str = "";
pub const ACCELERATE: &str = "accelerate";
pub const ACL: &str = "acl";
pub const ANALYTICS: &str = "analytics";
pub const CORS: &str = "cors";
pub const DELETE: &str = "delete";
pub const ENCRYPTION: &str = "encryption";
pub const INTELLIGENT_TIERING: &str = "intelligent-tiering";
pub const INVENTORY: &str = "inventory";
pub const LEGAL_HOLD: &str = "legal-hold";
pub const LIFECYCLE: &str = "lifecycle";
pub const LOCATION: &str = "location";
pub const LOGGING: &str = "logging";
pub const METRICS: &str = "metrics";
pub const NOTIFICATION: &str = "notification";
pub const OBJECT_LOCK: &str = "object-lock";
pub const OWNERSHIP_CONTROLS: &str = "ownershipControls";
pub const POLICY: &str = "policy";
pub const POLICY_STATUS: &str = "policyStatus";
pub const PUBLIC_ACCESS_BLOCK: &str = "publicAccessBlock";
pub const REPLICATION: &str = "replication";
pub const REQUEST_PAYMENT: &str = "requestPayment";
pub const RESTORE: &str = "restore";
pub const RETENTION: &str = "retention";
pub const SELECT: &str = "select";
pub const TAGGING: &str = "tagging";
pub const TORRENT: &str = "torrent";
pub const UPLOADS: &str = "uploads";
pub const VERSIONING: &str = "versioning";
pub const VERSIONS: &str = "versions";
pub const WEBSITE: &str = "website";
keywords: [
"accelerate" => ACCELERATE,
"acl" => ACL,
"analytics" => ANALYTICS,
"cors" => CORS,
"delete" => DELETE,
"encryption" => ENCRYPTION,
"intelligent-tiering" => INTELLIGENT_TIERING,
"inventory" => INVENTORY,
"legal-hold" => LEGAL_HOLD,
"lifecycle" => LIFECYCLE,
"location" => LOCATION,
"logging" => LOGGING,
"metrics" => METRICS,
"notification" => NOTIFICATION,
"object-lock" => OBJECT_LOCK,
"ownershipControls" => OWNERSHIP_CONTROLS,
"policy" => POLICY,
"policyStatus" => POLICY_STATUS,
"publicAccessBlock" => PUBLIC_ACCESS_BLOCK,
"replication" => REPLICATION,
"requestPayment" => REQUEST_PAYMENT,
"restore" => RESTORE,
"retention" => RETENTION,
"select" => SELECT,
"tagging" => TAGGING,
"torrent" => TORRENT,
"uploads" => UPLOADS,
"versioning" => VERSIONING,
"versions" => VERSIONS,
"website" => WEBSITE
],
fields: [
"continuation-token" => continuation_token,
"delimiter" => delimiter,
"encoding-type" => encoding_type,
"fetch-owner" => fetch_owner,
"id" => id,
"key-marker" => key_marker,
"list-type" => list_type,
"marker" => marker,
"max-keys" => max_keys,
"max-parts" => max_parts,
"max-uploads" => max_uploads,
"partNumber" => part_number,
"part-number-marker" => part_number_marker,
"prefix" => prefix,
"select-type" => select_type,
"start-after" => start_after,
"uploadId" => upload_id,
"upload-id-marker" => upload_id_marker,
"versionId" => version_id,
"version-id-marker" => version_id_marker
]
}
#[cfg(test)]
+5 -6
View File
@@ -1,6 +1,6 @@
[package]
name = "garage_block"
version = "0.8.0"
version = "0.8.1"
authors = ["Alex Auvolat <alex@adnab.me>"]
edition = "2018"
license = "AGPL-3.0"
@@ -14,10 +14,10 @@ path = "lib.rs"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
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" }
garage_db = { version = "0.8.1", path = "../db" }
garage_rpc = { version = "0.8.1", path = "../rpc" }
garage_util = { version = "0.8.1", path = "../util" }
garage_table = { version = "0.8.1", path = "../table" }
opentelemetry = "0.17"
@@ -31,7 +31,6 @@ rand = "0.8"
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"] }
serde_bytes = "0.11"
+116 -15
View File
@@ -3,8 +3,10 @@ use std::pin::Pin;
use std::sync::Arc;
use std::time::Duration;
use arc_swap::ArcSwapOption;
use async_trait::async_trait;
use bytes::Bytes;
use rand::prelude::*;
use serde::{Deserialize, Serialize};
use futures::Stream;
@@ -22,9 +24,12 @@ use garage_rpc::rpc_helper::netapp::stream::{stream_asyncread, ByteStream};
use garage_db as db;
use garage_util::background::{vars, BackgroundRunner};
use garage_util::data::*;
use garage_util::error::*;
use garage_util::metrics::RecordDuration;
use garage_util::persister::PersisterShared;
use garage_util::time::msec_to_rfc3339;
use garage_rpc::rpc_helper::OrderTag;
use garage_rpc::system::System;
@@ -87,7 +92,17 @@ pub struct BlockManager {
pub(crate) metrics: BlockManagerMetrics,
tx_scrub_command: mpsc::Sender<ScrubWorkerCommand>,
pub scrub_persister: PersisterShared<ScrubWorkerPersisted>,
tx_scrub_command: ArcSwapOption<mpsc::Sender<ScrubWorkerCommand>>,
}
#[derive(Serialize, Deserialize, Clone, Debug)]
pub struct BlockResyncErrorInfo {
pub hash: Hash,
pub refcount: u64,
pub error_count: u64,
pub last_try: u64,
pub next_try: u64,
}
// This custom struct contains functions that must only be ran
@@ -114,9 +129,10 @@ impl BlockManager {
.netapp
.endpoint("garage_block/manager.rs/Rpc".to_string());
let metrics = BlockManagerMetrics::new(resync.queue.clone(), resync.errors.clone());
let metrics =
BlockManagerMetrics::new(rc.rc.clone(), resync.queue.clone(), resync.errors.clone());
let (scrub_tx, scrub_rx) = mpsc::channel(1);
let scrub_persister = PersisterShared::new(&system.metadata_dir, "scrub_info");
let block_manager = Arc::new(Self {
replication,
@@ -128,21 +144,46 @@ impl BlockManager {
system,
endpoint,
metrics,
tx_scrub_command: scrub_tx,
scrub_persister,
tx_scrub_command: ArcSwapOption::new(None),
});
block_manager.endpoint.set_handler(block_manager.clone());
block_manager
}
pub fn spawn_workers(self: &Arc<Self>, bg: &BackgroundRunner) {
// Spawn a bunch of resync workers
for index in 0..MAX_RESYNC_WORKERS {
let worker = ResyncWorker::new(index, block_manager.clone());
block_manager.system.background.spawn_worker(worker);
let worker = ResyncWorker::new(index, self.clone());
bg.spawn_worker(worker);
}
// Spawn scrub worker
let scrub_worker = ScrubWorker::new(block_manager.clone(), scrub_rx);
block_manager.system.background.spawn_worker(scrub_worker);
let (scrub_tx, scrub_rx) = mpsc::channel(1);
self.tx_scrub_command.store(Some(Arc::new(scrub_tx)));
bg.spawn_worker(ScrubWorker::new(
self.clone(),
scrub_rx,
self.scrub_persister.clone(),
));
}
block_manager
pub fn register_bg_vars(&self, vars: &mut vars::BgVars) {
self.resync.register_bg_vars(vars);
vars.register_rw(
&self.scrub_persister,
"scrub-tranquility",
|p| p.get_with(|x| x.tranquility),
|p, tranquility| p.set_with(|x| x.tranquility = tranquility),
);
vars.register_ro(&self.scrub_persister, "scrub-last-completed", |p| {
p.get_with(|x| msec_to_rfc3339(x.time_last_complete_scrub))
});
vars.register_ro(&self.scrub_persister, "scrub-corruptions_detected", |p| {
p.get_with(|x| x.corruptions_detected)
});
}
/// Ask nodes that might have a (possibly compressed) block for it
@@ -309,9 +350,42 @@ impl BlockManager {
Ok(self.rc.rc.len()?)
}
/// Get number of items in the refcount table
pub fn rc_fast_len(&self) -> Result<Option<usize>, Error> {
Ok(self.rc.rc.fast_len()?)
}
/// Send command to start/stop/manager scrub worker
pub async fn send_scrub_command(&self, cmd: ScrubWorkerCommand) {
let _ = self.tx_scrub_command.send(cmd).await;
pub async fn send_scrub_command(&self, cmd: ScrubWorkerCommand) -> Result<(), Error> {
let tx = self.tx_scrub_command.load();
let tx = tx.as_ref().ok_or_message("scrub worker is not running")?;
tx.send(cmd).await.ok_or_message("send error")?;
Ok(())
}
/// Get the reference count of a block
pub fn get_block_rc(&self, hash: &Hash) -> Result<u64, Error> {
Ok(self.rc.get_block_rc(hash)?.as_u64())
}
/// List all resync errors
pub fn list_resync_errors(&self) -> Result<Vec<BlockResyncErrorInfo>, Error> {
let mut blocks = Vec::with_capacity(self.resync.errors.len());
for ent in self.resync.errors.iter()? {
let (hash, cnt) = ent?;
let cnt = ErrorCounter::decode(&cnt);
blocks.push(BlockResyncErrorInfo {
hash: Hash::try_from(&hash).unwrap(),
refcount: 0,
error_count: cnt.errors,
last_try: cnt.last_try,
next_try: cnt.next_try(),
});
}
for block in blocks.iter_mut() {
block.refcount = self.get_block_rc(&block.hash)?;
}
Ok(blocks)
}
//// ----- Managing the reference counter ----
@@ -603,14 +677,21 @@ impl BlockManagerLocked {
}
};
let mut path2 = path.clone();
path2.set_extension("tmp");
let mut f = fs::File::create(&path2).await?;
let mut path_tmp = path.clone();
let tmp_extension = format!("tmp{}", hex::encode(thread_rng().gen::<[u8; 4]>()));
path_tmp.set_extension(tmp_extension);
let mut delete_on_drop = DeleteOnDrop(Some(path_tmp.clone()));
let mut f = fs::File::create(&path_tmp).await?;
f.write_all(data).await?;
f.sync_all().await?;
drop(f);
fs::rename(path2, path).await?;
fs::rename(path_tmp, path).await?;
delete_on_drop.cancel();
if let Some(to_delete) = to_delete {
fs::remove_file(to_delete).await?;
}
@@ -676,3 +757,23 @@ async fn read_stream_to_end(mut stream: ByteStream) -> Result<Bytes, Error> {
.concat()
.into())
}
struct DeleteOnDrop(Option<PathBuf>);
impl DeleteOnDrop {
fn cancel(&mut self) {
drop(self.0.take());
}
}
impl Drop for DeleteOnDrop {
fn drop(&mut self) {
if let Some(path) = self.0.take() {
tokio::spawn(async move {
if let Err(e) = fs::remove_file(&path).await {
debug!("DeleteOnDrop failed for {}: {}", path.display(), e);
}
});
}
}
}
+11 -1
View File
@@ -1,9 +1,11 @@
use opentelemetry::{global, metrics::*};
use garage_db as db;
use garage_db::counted_tree_hack::CountedTree;
/// TableMetrics reference all counter used for metrics
pub struct BlockManagerMetrics {
pub(crate) _rc_size: ValueObserver<u64>,
pub(crate) _resync_queue_len: ValueObserver<u64>,
pub(crate) _resync_errored_blocks: ValueObserver<u64>,
@@ -23,9 +25,17 @@ pub struct BlockManagerMetrics {
}
impl BlockManagerMetrics {
pub fn new(resync_queue: CountedTree, resync_errors: CountedTree) -> Self {
pub fn new(rc_tree: db::Tree, resync_queue: CountedTree, resync_errors: CountedTree) -> Self {
let meter = global::meter("garage_model/block");
Self {
_rc_size: meter
.u64_value_observer("block.rc_size", move |observer| {
if let Ok(Some(v)) = rc_tree.fast_len() {
observer.observe(v as u64, &[])
}
})
.with_description("Number of blocks known to the reference counter")
.init(),
_resync_queue_len: meter
.u64_value_observer("block.resync_queue_length", move |observer| {
observer.observe(resync_queue.len() as u64, &[])
+7
View File
@@ -169,4 +169,11 @@ impl RcEntry {
pub(crate) fn is_needed(&self) -> bool {
!self.is_deletable()
}
pub(crate) fn as_u64(&self) -> u64 {
match self {
RcEntry::Present { count } => *count,
_ => 0,
}
}
}
+72 -57
View File
@@ -13,7 +13,7 @@ use tokio::sync::watch;
use garage_util::background::*;
use garage_util::data::*;
use garage_util::error::*;
use garage_util::persister::Persister;
use garage_util::persister::PersisterShared;
use garage_util::time::*;
use garage_util::tranquilizer::Tranquilizer;
@@ -53,7 +53,7 @@ impl Worker for RepairWorker {
"Block repair worker".into()
}
fn info(&self) -> Option<String> {
fn status(&self) -> WorkerStatus {
match self.block_iter.as_ref() {
None => {
let idx_bytes = self
@@ -66,9 +66,20 @@ impl Worker for RepairWorker {
} else {
idx_bytes
};
Some(format!("Phase 1: {}", hex::encode(idx_bytes)))
WorkerStatus {
progress: Some("0.00%".into()),
freeform: vec![format!(
"Currently in phase 1, iterator position: {}",
hex::encode(idx_bytes)
)],
..Default::default()
}
}
Some(bi) => Some(format!("Phase 2: {:.2}% done", bi.progress() * 100.)),
Some(bi) => WorkerStatus {
progress: Some(format!("{:.2}%", bi.progress() * 100.)),
freeform: vec!["Currently in phase 2".into()],
..Default::default()
},
}
}
@@ -137,7 +148,7 @@ impl Worker for RepairWorker {
}
}
async fn wait_for_work(&mut self, _must_exit: &watch::Receiver<bool>) -> WorkerState {
async fn wait_for_work(&mut self) -> WorkerState {
unreachable!()
}
}
@@ -157,15 +168,24 @@ pub struct ScrubWorker {
work: ScrubWorkerState,
tranquilizer: Tranquilizer,
persister: Persister<ScrubWorkerPersisted>,
persisted: ScrubWorkerPersisted,
persister: PersisterShared<ScrubWorkerPersisted>,
}
#[derive(Serialize, Deserialize)]
struct ScrubWorkerPersisted {
tranquility: u32,
time_last_complete_scrub: u64,
corruptions_detected: u64,
pub struct ScrubWorkerPersisted {
pub tranquility: u32,
pub(crate) time_last_complete_scrub: u64,
pub(crate) corruptions_detected: u64,
}
impl garage_util::migrate::InitialFormat for ScrubWorkerPersisted {}
impl Default for ScrubWorkerPersisted {
fn default() -> Self {
ScrubWorkerPersisted {
time_last_complete_scrub: 0,
tranquility: INITIAL_SCRUB_TRANQUILITY,
corruptions_detected: 0,
}
}
}
enum ScrubWorkerState {
@@ -186,27 +206,20 @@ pub enum ScrubWorkerCommand {
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,
},
};
pub(crate) fn new(
manager: Arc<BlockManager>,
rx_cmd: mpsc::Receiver<ScrubWorkerCommand>,
persister: PersisterShared<ScrubWorkerPersisted>,
) -> Self {
Self {
manager,
rx_cmd,
work: ScrubWorkerState::Finished,
tranquilizer: Tranquilizer::new(30),
persister,
persisted,
}
}
@@ -255,12 +268,6 @@ impl ScrubWorker {
}
}
}
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);
}
}
}
}
}
@@ -271,29 +278,37 @@ impl Worker for ScrubWorker {
"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)
fn status(&self) -> WorkerStatus {
let (corruptions_detected, tranquility, time_last_complete_scrub) =
self.persister.get_with(|p| {
(
p.corruptions_detected,
p.tranquility,
p.time_last_complete_scrub,
)
}
ScrubWorkerState::Finished => format!(
"Last completed scrub: {}",
msec_to_rfc3339(self.persisted.time_last_complete_scrub)
),
});
let mut s = WorkerStatus {
persistent_errors: Some(corruptions_detected),
tranquility: Some(tranquility),
..Default::default()
};
Some(format!(
"{} ; corruptions detected: {}",
s, self.persisted.corruptions_detected
))
match &self.work {
ScrubWorkerState::Running(bsi) => {
s.progress = Some(format!("{:.2}%", bsi.progress() * 100.));
}
ScrubWorkerState::Paused(bsi, rt) => {
s.progress = Some(format!("{:.2}%", bsi.progress() * 100.));
s.freeform = vec![format!("Scrub paused, resumes at {}", msec_to_rfc3339(*rt))];
}
ScrubWorkerState::Finished => {
s.freeform = vec![format!(
"Last scrub completed at {}",
msec_to_rfc3339(time_last_complete_scrub)
)];
}
}
s
}
async fn work(&mut self, _must_exit: &mut watch::Receiver<bool>) -> Result<WorkerState, Error> {
@@ -310,18 +325,17 @@ impl Worker for ScrubWorker {
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?;
self.persister.set_with(|p| p.corruptions_detected += 1)?;
}
Err(e) => return Err(e),
_ => (),
};
Ok(self
.tranquilizer
.tranquilize_worker(self.persisted.tranquility))
.tranquilize_worker(self.persister.get_with(|p| p.tranquility)))
} else {
self.persisted.time_last_complete_scrub = now_msec();
self.persister.save_async(&self.persisted).await?;
self.persister
.set_with(|p| p.time_last_complete_scrub = now_msec())?;
self.work = ScrubWorkerState::Finished;
self.tranquilizer.clear();
Ok(WorkerState::Idle)
@@ -331,12 +345,13 @@ impl Worker for ScrubWorker {
}
}
async fn wait_for_work(&mut self, _must_exit: &watch::Receiver<bool>) -> WorkerState {
async fn wait_for_work(&mut self) -> 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,
self.persister.get_with(|p| p.time_last_complete_scrub)
+ SCRUB_INTERVAL.as_millis() as u64,
ScrubWorkerCommand::Start,
),
};
+95 -71
View File
@@ -3,7 +3,6 @@ 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};
@@ -22,7 +21,7 @@ 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::persister::PersisterShared;
use garage_util::time::*;
use garage_util::tranquilizer::Tranquilizer;
@@ -49,13 +48,12 @@ const INITIAL_RESYNC_TRANQUILITY: u32 = 2;
pub struct BlockResyncManager {
pub(crate) queue: CountedTree,
pub(crate) notify: Notify,
pub(crate) notify: Arc<Notify>,
pub(crate) errors: CountedTree,
busy_set: BusySet,
persister: Persister<ResyncPersistedConfig>,
persisted: ArcSwap<ResyncPersistedConfig>,
persister: PersisterShared<ResyncPersistedConfig>,
}
#[derive(Serialize, Deserialize, Clone, Copy)]
@@ -63,6 +61,15 @@ struct ResyncPersistedConfig {
n_workers: usize,
tranquility: u32,
}
impl garage_util::migrate::InitialFormat for ResyncPersistedConfig {}
impl Default for ResyncPersistedConfig {
fn default() -> Self {
ResyncPersistedConfig {
n_workers: 1,
tranquility: INITIAL_RESYNC_TRANQUILITY,
}
}
}
enum ResyncIterResult {
BusyDidSomething,
@@ -90,22 +97,14 @@ impl BlockResyncManager {
.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,
},
};
let persister = PersisterShared::new(&system.metadata_dir, "resync_cfg");
Self {
queue,
notify: Notify::new(),
notify: Arc::new(Notify::new()),
errors,
busy_set: Arc::new(Mutex::new(HashSet::new())),
persister,
persisted: ArcSwap::new(Arc::new(persisted)),
}
}
@@ -123,6 +122,56 @@ impl BlockResyncManager {
Ok(self.errors.len())
}
/// Clear the error counter for a block and put it in queue immediately
pub fn clear_backoff(&self, hash: &Hash) -> Result<(), Error> {
let now = now_msec();
if let Some(ec) = self.errors.get(hash)? {
let mut ec = ErrorCounter::decode(&ec);
if ec.errors > 0 {
ec.last_try = now - ec.delay_msec();
self.errors.insert(hash, ec.encode())?;
self.put_to_resync_at(hash, now)?;
return Ok(());
}
}
Err(Error::Message(format!(
"Block {:?} was not in an errored state",
hash
)))
}
pub fn register_bg_vars(&self, vars: &mut vars::BgVars) {
let notify = self.notify.clone();
vars.register_rw(
&self.persister,
"resync-worker-count",
|p| p.get_with(|x| x.n_workers),
move |p, n_workers| {
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
)));
}
p.set_with(|x| x.n_workers = n_workers)?;
notify.notify_waiters();
Ok(())
},
);
let notify = self.notify.clone();
vars.register_rw(
&self.persister,
"resync-tranquility",
|p| p.get_with(|x| x.tranquility),
move |p, tranquility| {
p.set_with(|x| x.tranquility = tranquility)?;
notify.notify_waiters();
Ok(())
},
);
}
// ---- Resync loop ----
// This part manages a queue of blocks that need to be
@@ -257,7 +306,7 @@ impl BlockResyncManager {
if let Err(e) = &res {
manager.metrics.resync_error_counter.add(1);
warn!("Error when resyncing {:?}: {}", hash, e);
error!("Error when resyncing {:?}: {}", hash, e);
let err_counter = match self.errors.get(hash.as_slice())? {
Some(ec) => ErrorCounter::decode(&ec).add1(now + 1),
@@ -417,33 +466,6 @@ impl BlockResyncManager {
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 {
@@ -458,15 +480,18 @@ pub(crate) struct ResyncWorker {
manager: Arc<BlockManager>,
tranquilizer: Tranquilizer,
next_delay: Duration,
persister: PersisterShared<ResyncPersistedConfig>,
}
impl ResyncWorker {
pub(crate) fn new(index: usize, manager: Arc<BlockManager>) -> Self {
let persister = manager.resync.persister.clone();
Self {
index,
manager,
tranquilizer: Tranquilizer::new(30),
next_delay: Duration::from_secs(10),
persister,
}
}
}
@@ -477,39 +502,36 @@ impl Worker for ResyncWorker {
format!("Block resync worker #{}", self.index + 1)
}
fn info(&self) -> Option<String> {
let persisted = self.manager.resync.persisted.load();
fn status(&self) -> WorkerStatus {
let (n_workers, tranquility) = self.persister.get_with(|x| (x.n_workers, x.tranquility));
if self.index >= persisted.n_workers {
return Some("(unused)".into());
if self.index >= n_workers {
return WorkerStatus {
freeform: vec!["This worker is currently disabled".into()],
..Default::default()
};
}
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));
WorkerStatus {
queue_length: Some(self.manager.resync.queue_len().unwrap_or(0) as u64),
tranquility: Some(tranquility),
persistent_errors: Some(self.manager.resync.errors_len().unwrap_or(0) as u64),
..Default::default()
}
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 {
let (n_workers, tranquility) = self.persister.get_with(|x| (x.n_workers, x.tranquility));
if self.index >= 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::BusyDidSomething) => {
Ok(self.tranquilizer.tranquilize_worker(tranquility))
}
Ok(ResyncIterResult::BusyDidNothing) => Ok(WorkerState::Busy),
Ok(ResyncIterResult::IdleFor(delay)) => {
self.next_delay = delay;
@@ -527,8 +549,8 @@ impl Worker for ResyncWorker {
}
}
async fn wait_for_work(&mut self, _must_exit: &watch::Receiver<bool>) -> WorkerState {
while self.index >= self.manager.resync.persisted.load().n_workers {
async fn wait_for_work(&mut self) -> WorkerState {
while self.index >= self.persister.get_with(|x| x.n_workers) {
self.manager.resync.notify.notified().await
}
@@ -545,9 +567,9 @@ impl Worker for ResyncWorker {
/// and the time of the last try.
/// Used to implement exponential backoff.
#[derive(Clone, Copy, Debug)]
struct ErrorCounter {
errors: u64,
last_try: u64,
pub(crate) struct ErrorCounter {
pub(crate) errors: u64,
pub(crate) last_try: u64,
}
impl ErrorCounter {
@@ -558,12 +580,13 @@ impl ErrorCounter {
}
}
fn decode(data: &[u8]) -> Self {
pub(crate) 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),
@@ -583,7 +606,8 @@ impl ErrorCounter {
(RESYNC_RETRY_DELAY.as_millis() as u64)
<< std::cmp::min(self.errors - 1, RESYNC_RETRY_DELAY_MAX_BACKOFF_POWER)
}
fn next_try(&self) -> u64 {
pub(crate) fn next_try(&self) -> u64 {
self.last_try + self.delay_msec()
}
}
+1 -1
View File
@@ -1,6 +1,6 @@
[package]
name = "garage_db"
version = "0.8.0"
version = "0.8.1"
authors = ["Alex Auvolat <alex@adnab.me>"]
edition = "2018"
license = "AGPL-3.0"
+7
View File
@@ -181,6 +181,10 @@ impl Tree {
pub fn len(&self) -> Result<usize> {
self.0.len(self.1)
}
#[inline]
pub fn fast_len(&self) -> Result<Option<usize>> {
self.0.fast_len(self.1)
}
#[inline]
pub fn first(&self) -> Result<Option<(Value, Value)>> {
@@ -323,6 +327,9 @@ pub(crate) trait IDb: Send + Sync {
fn get(&self, tree: usize, key: &[u8]) -> Result<Option<Value>>;
fn len(&self, tree: usize) -> Result<usize>;
fn fast_len(&self, _tree: usize) -> Result<Option<usize>> {
Ok(None)
}
fn insert(&self, tree: usize, key: &[u8], value: &[u8]) -> Result<Option<Value>>;
fn remove(&self, tree: usize, key: &[u8]) -> Result<Option<Value>>;
+4
View File
@@ -121,6 +121,10 @@ impl IDb for LmdbDb {
Ok(tree.len(&tx)?.try_into().unwrap())
}
fn fast_len(&self, tree: usize) -> Result<Option<usize>> {
Ok(Some(self.len(tree)?))
}
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()?;
+4
View File
@@ -144,6 +144,10 @@ impl IDb for SqliteDb {
}
}
fn fast_len(&self, tree: usize) -> Result<Option<usize>> {
Ok(Some(self.len(tree)?))
}
fn insert(&self, tree: usize, key: &[u8], value: &[u8]) -> Result<Option<Value>> {
trace!("insert {}: lock db", tree);
let this = self.0.lock().unwrap();
+11 -12
View File
@@ -1,6 +1,6 @@
[package]
name = "garage"
version = "0.8.0"
version = "0.8.1"
authors = ["Alex Auvolat <alex@adnab.me>"]
edition = "2018"
license = "AGPL-3.0"
@@ -21,14 +21,14 @@ path = "tests/lib.rs"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
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" }
garage_db = { version = "0.8.1", path = "../db" }
garage_api = { version = "0.8.1", path = "../api" }
garage_block = { version = "0.8.1", path = "../block" }
garage_model = { version = "0.8.1", path = "../model" }
garage_rpc = { version = "0.8.1", path = "../rpc" }
garage_table = { version = "0.8.1", path = "../table" }
garage_util = { version = "0.8.1", path = "../util" }
garage_web = { version = "0.8.1", path = "../web" }
backtrace = "0.3"
bytes = "1.0"
@@ -36,13 +36,12 @@ bytesize = "1.1"
timeago = "0.3"
parse_duration = "2.1"
hex = "0.4"
tracing = { version = "0.1.30", features = ["log-always"] }
tracing = { version = "0.1.30" }
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" }
rmp-serde = "0.15"
serde = { version = "1.0", default-features = false, features = ["derive", "rc"] }
serde_bytes = "0.11"
structopt = { version = "0.3", default-features = false }
@@ -74,7 +73,7 @@ base64 = "0.13"
[features]
default = [ "bundled-libs", "metrics", "sled" ]
default = [ "bundled-libs", "metrics", "sled", "k2v" ]
k2v = [ "garage_util/k2v", "garage_api/k2v" ]
+350 -76
View File
@@ -5,9 +5,11 @@ use std::sync::Arc;
use async_trait::async_trait;
use serde::{Deserialize, Serialize};
use garage_util::background::BackgroundRunner;
use garage_util::crdt::*;
use garage_util::data::*;
use garage_util::error::Error as GarageError;
use garage_util::formater::format_table_to_string;
use garage_util::time::*;
use garage_table::replication::*;
@@ -15,7 +17,7 @@ use garage_table::*;
use garage_rpc::*;
use garage_block::repair::ScrubWorkerCommand;
use garage_block::manager::BlockResyncErrorInfo;
use garage_model::bucket_alias_table::*;
use garage_model::bucket_table::*;
@@ -24,6 +26,8 @@ use garage_model::helper::error::{Error, OkOrBadRequest};
use garage_model::key_table::*;
use garage_model::migrate::Migrate;
use garage_model::permission::*;
use garage_model::s3::object_table::*;
use garage_model::s3::version_table::Version;
use crate::cli::*;
use crate::repair::online::launch_online_repair;
@@ -38,7 +42,8 @@ pub enum AdminRpc {
LaunchRepair(RepairOpt),
Migrate(MigrateOpt),
Stats(StatsOpt),
Worker(WorkerOpt),
Worker(WorkerOperation),
BlockOperation(BlockOperation),
// Replies
Ok(String),
@@ -54,6 +59,14 @@ pub enum AdminRpc {
HashMap<usize, garage_util::background::WorkerInfo>,
WorkerListOpt,
),
WorkerVars(Vec<(Uuid, String, String)>),
WorkerInfo(usize, garage_util::background::WorkerInfo),
BlockErrorList(Vec<BlockResyncErrorInfo>),
BlockInfo {
hash: Hash,
refcount: u64,
versions: Vec<Result<Version, Uuid>>,
},
}
impl Rpc for AdminRpc {
@@ -62,17 +75,24 @@ impl Rpc for AdminRpc {
pub struct AdminRpcHandler {
garage: Arc<Garage>,
background: Arc<BackgroundRunner>,
endpoint: Arc<Endpoint<AdminRpc, Self>>,
}
impl AdminRpcHandler {
pub fn new(garage: Arc<Garage>) -> Arc<Self> {
pub fn new(garage: Arc<Garage>, background: Arc<BackgroundRunner>) -> Arc<Self> {
let endpoint = garage.system.netapp.endpoint(ADMIN_RPC_PATH.into());
let admin = Arc::new(Self { garage, endpoint });
let admin = Arc::new(Self {
garage,
background,
endpoint,
});
admin.endpoint.set_handler(admin.clone());
admin
}
// ================ BUCKET COMMANDS ====================
async fn handle_bucket_cmd(&self, cmd: &BucketOperation) -> Result<AdminRpc, Error> {
match cmd {
BucketOperation::List => self.handle_list_buckets().await,
@@ -551,6 +571,8 @@ impl AdminRpcHandler {
Ok(AdminRpc::Ok(ret))
}
// ================ KEY COMMANDS ====================
async fn handle_key_cmd(&self, cmd: &KeyOperation) -> Result<AdminRpc, Error> {
match cmd {
KeyOperation::List => self.handle_list_keys().await,
@@ -688,6 +710,8 @@ impl AdminRpcHandler {
Ok(AdminRpc::KeyInfo(key, relevant_buckets))
}
// ================ MIGRATION COMMANDS ====================
async fn handle_migrate(self: &Arc<Self>, opt: MigrateOpt) -> Result<AdminRpc, Error> {
if !opt.yes {
return Err(Error::BadRequest(
@@ -704,6 +728,8 @@ impl AdminRpcHandler {
Ok(AdminRpc::Ok("Migration successfull.".into()))
}
// ================ REPAIR COMMANDS ====================
async fn handle_launch_repair(self: &Arc<Self>, opt: RepairOpt) -> Result<AdminRpc, Error> {
if !opt.yes {
return Err(Error::BadRequest(
@@ -739,7 +765,7 @@ impl AdminRpcHandler {
)))
}
} else {
launch_online_repair(self.garage.clone(), opt).await;
launch_online_repair(&self.garage, &self.background, opt).await?;
Ok(AdminRpc::Ok(format!(
"Repair launched on {:?}",
self.garage.system.id
@@ -747,6 +773,8 @@ impl AdminRpcHandler {
}
}
// ================ STATS COMMANDS ====================
async fn handle_stats(&self, opt: StatsOpt) -> Result<AdminRpc, Error> {
if opt.all_nodes {
let mut ret = String::new();
@@ -763,11 +791,12 @@ impl AdminRpcHandler {
match self
.endpoint
.call(&node_id, AdminRpc::Stats(opt), PRIO_NORMAL)
.await?
.await
{
Ok(AdminRpc::Ok(s)) => writeln!(&mut ret, "{}", s).unwrap(),
Ok(x) => writeln!(&mut ret, "Bad answer: {:?}", x).unwrap(),
Err(e) => writeln!(&mut ret, "Error: {}", e).unwrap(),
Ok(Ok(AdminRpc::Ok(s))) => writeln!(&mut ret, "{}", s).unwrap(),
Ok(Ok(x)) => writeln!(&mut ret, "Bad answer: {:?}", x).unwrap(),
Ok(Err(e)) => writeln!(&mut ret, "Remote error: {}", e).unwrap(),
Err(e) => writeln!(&mut ret, "Network error: {}", e).unwrap(),
}
}
Ok(AdminRpc::Ok(ret))
@@ -787,6 +816,7 @@ impl AdminRpcHandler {
.unwrap_or_else(|| "(unknown)".into()),
)
.unwrap();
writeln!(&mut ret, "\nDatabase engine: {}", self.garage.db.engine()).unwrap();
// Gather ring statistics
@@ -805,21 +835,38 @@ 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)?;
// Gather table statistics
let mut table = vec![" Table\tItems\tMklItems\tMklTodo\tGcTodo".into()];
table.push(self.gather_table_stats(&self.garage.bucket_table, opt.detailed)?);
table.push(self.gather_table_stats(&self.garage.key_table, opt.detailed)?);
table.push(self.gather_table_stats(&self.garage.object_table, opt.detailed)?);
table.push(self.gather_table_stats(&self.garage.version_table, opt.detailed)?);
table.push(self.gather_table_stats(&self.garage.block_ref_table, opt.detailed)?);
write!(
&mut ret,
"\nTable stats:\n{}",
format_table_to_string(table)
)
.unwrap();
// Gather block manager statistics
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()?
)
.unwrap();
}
let rc_len = if opt.detailed {
self.garage.block_manager.rc_len()?.to_string()
} else {
self.garage
.block_manager
.rc_fast_len()?
.map(|x| x.to_string())
.unwrap_or_else(|| "NC".into())
};
writeln!(
&mut ret,
" number of RC entries (~= number of blocks): {}",
rc_len
)
.unwrap();
writeln!(
&mut ret,
" resync queue length: {}",
@@ -833,79 +880,305 @@ impl AdminRpcHandler {
)
.unwrap();
if !opt.detailed {
writeln!(&mut ret, "\nIf values are missing (marked as NC), consider adding the --detailed flag - this will be slow.").unwrap();
}
Ok(ret)
}
fn gather_table_stats<F, R>(
&self,
to: &mut String,
t: &Arc<Table<F, R>>,
opt: &StatsOpt,
) -> Result<(), Error>
detailed: bool,
) -> Result<String, 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().map_err(GarageError::from)?
let (data_len, mkl_len) = if detailed {
(
t.data.store.len().map_err(GarageError::from)?.to_string(),
t.merkle_updater.merkle_tree_len()?.to_string(),
)
.unwrap();
writeln!(
to,
" Merkle tree size: {}",
t.merkle_updater.merkle_tree_len()?
} else {
(
t.data
.store
.fast_len()
.map_err(GarageError::from)?
.map(|x| x.to_string())
.unwrap_or_else(|| "NC".into()),
t.merkle_updater
.merkle_tree_fast_len()?
.map(|x| x.to_string())
.unwrap_or_else(|| "NC".into()),
)
.unwrap();
}
writeln!(
to,
" Merkle updater todo queue length: {}",
t.merkle_updater.todo_len()?
)
.unwrap();
writeln!(to, " GC todo queue length: {}", t.data.gc_todo_len()?).unwrap();
};
Ok(())
Ok(format!(
" {}\t{}\t{}\t{}\t{}",
F::TABLE_NAME,
data_len,
mkl_len,
t.merkle_updater.todo_len()?,
t.data.gc_todo_len()?
))
}
// ----
// ================ WORKER COMMANDS ====================
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))
async fn handle_worker_cmd(&self, cmd: &WorkerOperation) -> Result<AdminRpc, Error> {
match cmd {
WorkerOperation::List { opt } => {
let workers = self.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()))
WorkerOperation::Info { tid } => {
let info = self
.background
.get_worker_info()
.get(tid)
.ok_or_bad_request(format!("No worker with TID {}", tid))?
.clone();
Ok(AdminRpc::WorkerInfo(*tid, info))
}
WorkerOperation::Get {
all_nodes,
variable,
} => self.handle_get_var(*all_nodes, variable).await,
WorkerOperation::Set {
all_nodes,
variable,
value,
} => self.handle_set_var(*all_nodes, variable, value).await,
}
}
async fn handle_get_var(
&self,
all_nodes: bool,
variable: &Option<String>,
) -> Result<AdminRpc, Error> {
if all_nodes {
let mut ret = vec![];
let ring = self.garage.system.ring.borrow().clone();
for node in ring.layout.node_ids().iter() {
let node = (*node).into();
match self
.endpoint
.call(
&node,
AdminRpc::Worker(WorkerOperation::Get {
all_nodes: false,
variable: variable.clone(),
}),
PRIO_NORMAL,
)
.await??
{
AdminRpc::WorkerVars(v) => ret.extend(v),
m => return Err(GarageError::unexpected_rpc_message(m).into()),
}
WorkerSetCmd::ResyncNWorkers { n_workers } => {
self.garage
.block_manager
.resync
.set_n_workers(n_workers)
}
Ok(AdminRpc::WorkerVars(ret))
} else {
#[allow(clippy::collapsible_else_if)]
if let Some(v) = variable {
Ok(AdminRpc::WorkerVars(vec![(
self.garage.system.id,
v.clone(),
self.garage.bg_vars.get(v)?,
)]))
} else {
let mut vars = self.garage.bg_vars.get_all();
vars.sort();
Ok(AdminRpc::WorkerVars(
vars.into_iter()
.map(|(k, v)| (self.garage.system.id, k.to_string(), v))
.collect(),
))
}
}
}
async fn handle_set_var(
&self,
all_nodes: bool,
variable: &str,
value: &str,
) -> Result<AdminRpc, Error> {
if all_nodes {
let mut ret = vec![];
let ring = self.garage.system.ring.borrow().clone();
for node in ring.layout.node_ids().iter() {
let node = (*node).into();
match self
.endpoint
.call(
&node,
AdminRpc::Worker(WorkerOperation::Set {
all_nodes: false,
variable: variable.to_string(),
value: value.to_string(),
}),
PRIO_NORMAL,
)
.await??
{
AdminRpc::WorkerVars(v) => ret.extend(v),
m => return Err(GarageError::unexpected_rpc_message(m).into()),
}
}
Ok(AdminRpc::WorkerVars(ret))
} else {
self.garage.bg_vars.set(variable, value)?;
Ok(AdminRpc::WorkerVars(vec![(
self.garage.system.id,
variable.to_string(),
value.to_string(),
)]))
}
}
// ================ BLOCK COMMANDS ====================
async fn handle_block_cmd(&self, cmd: &BlockOperation) -> Result<AdminRpc, Error> {
match cmd {
BlockOperation::ListErrors => Ok(AdminRpc::BlockErrorList(
self.garage.block_manager.list_resync_errors()?,
)),
BlockOperation::Info { hash } => {
let hash = hex::decode(hash).ok_or_bad_request("invalid hash")?;
let hash = Hash::try_from(&hash).ok_or_bad_request("invalid hash")?;
let refcount = self.garage.block_manager.get_block_rc(&hash)?;
let block_refs = self
.garage
.block_ref_table
.get_range(&hash, None, None, 10000, Default::default())
.await?;
let mut versions = vec![];
for br in block_refs {
if let Some(v) = self
.garage
.version_table
.get(&br.version, &EmptyKey)
.await?
{
versions.push(Ok(v));
} else {
versions.push(Err(br.version));
}
}
Ok(AdminRpc::BlockInfo {
hash,
refcount,
versions,
})
}
BlockOperation::RetryNow { all, blocks } => {
if *all {
if !blocks.is_empty() {
return Err(Error::BadRequest(
"--all was specified, cannot also specify blocks".into(),
));
}
let blocks = self.garage.block_manager.list_resync_errors()?;
for b in blocks.iter() {
self.garage.block_manager.resync.clear_backoff(&b.hash)?;
}
Ok(AdminRpc::Ok(format!(
"{} blocks returned in queue for a retry now (check logs to see results)",
blocks.len()
)))
} else {
for hash in blocks {
let hash = hex::decode(hash).ok_or_bad_request("invalid hash")?;
let hash = Hash::try_from(&hash).ok_or_bad_request("invalid hash")?;
self.garage.block_manager.resync.clear_backoff(&hash)?;
}
Ok(AdminRpc::Ok(format!(
"{} blocks returned in queue for a retry now (check logs to see results)",
blocks.len()
)))
}
}
BlockOperation::Purge { yes, blocks } => {
if !yes {
return Err(Error::BadRequest(
"Pass the --yes flag to confirm block purge operation.".into(),
));
}
let mut obj_dels = 0;
let mut ver_dels = 0;
for hash in blocks {
let hash = hex::decode(hash).ok_or_bad_request("invalid hash")?;
let hash = Hash::try_from(&hash).ok_or_bad_request("invalid hash")?;
let block_refs = self
.garage
.block_ref_table
.get_range(&hash, None, None, 10000, Default::default())
.await?;
Ok(AdminRpc::Ok("Number of resync workers updated".into()))
for br in block_refs {
let version = match self
.garage
.version_table
.get(&br.version, &EmptyKey)
.await?
{
Some(v) => v,
None => continue,
};
if let Some(object) = self
.garage
.object_table
.get(&version.bucket_id, &version.key)
.await?
{
let ov = object.versions().iter().rev().find(|v| v.is_complete());
if let Some(ov) = ov {
if ov.uuid == br.version {
let del_uuid = gen_uuid();
let deleted_object = Object::new(
version.bucket_id,
version.key.clone(),
vec![ObjectVersion {
uuid: del_uuid,
timestamp: ov.timestamp + 1,
state: ObjectVersionState::Complete(
ObjectVersionData::DeleteMarker,
),
}],
);
self.garage.object_table.insert(&deleted_object).await?;
obj_dels += 1;
}
}
}
if !version.deleted.get() {
let deleted_version = Version::new(
version.uuid,
version.bucket_id,
version.key.clone(),
true,
);
self.garage.version_table.insert(&deleted_version).await?;
ver_dels += 1;
}
}
}
WorkerSetCmd::ResyncTranquility { tranquility } => {
self.garage
.block_manager
.resync
.set_tranquility(tranquility)
.await?;
Ok(AdminRpc::Ok("Resync tranquility updated".into()))
}
},
Ok(AdminRpc::Ok(format!(
"{} blocks were purged: {} object deletion markers added, {} versions marked deleted",
blocks.len(),
obj_dels,
ver_dels
)))
}
}
}
}
@@ -923,7 +1196,8 @@ 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,
AdminRpc::Worker(wo) => self.handle_worker_cmd(wo).await,
AdminRpc::BlockOperation(bo) => self.handle_block_cmd(bo).await,
m => Err(GarageError::unexpected_rpc_message(m).into()),
}
}
+20 -1
View File
@@ -41,6 +41,9 @@ pub async fn cli_command_dispatch(
}
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,
Command::Block(bo) => {
cmd_admin(admin_rpc_endpoint, rpc_host, AdminRpc::BlockOperation(bo)).await
}
_ => unreachable!(),
}
}
@@ -186,7 +189,23 @@ pub async fn cmd_admin(
print_key_info(&key, &rb);
}
AdminRpc::WorkerList(wi, wlo) => {
print_worker_info(wi, wlo);
print_worker_list(wi, wlo);
}
AdminRpc::WorkerVars(wv) => {
print_worker_vars(wv);
}
AdminRpc::WorkerInfo(tid, wi) => {
print_worker_info(tid, wi);
}
AdminRpc::BlockErrorList(el) => {
print_block_error_list(el);
}
AdminRpc::BlockInfo {
hash,
refcount,
versions,
} => {
print_block_info(hash, refcount, versions);
}
r => {
error!("Unexpected response: {:?}", r);
+54 -20
View File
@@ -49,7 +49,11 @@ pub enum Command {
/// Manage background workers
#[structopt(name = "worker", version = garage_version())]
Worker(WorkerOpt),
Worker(WorkerOperation),
/// Low-level debug operations on data blocks
#[structopt(name = "block", version = garage_version())]
Block(BlockOperation),
}
#[derive(StructOpt, Debug)]
@@ -502,25 +506,36 @@ pub struct StatsOpt {
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 {
pub enum WorkerOperation {
/// List all workers on Garage node
#[structopt(name = "list", version = garage_version())]
List {
#[structopt(flatten)]
opt: WorkerListOpt,
},
/// Get detailed information about a worker
#[structopt(name = "info", version = garage_version())]
Info { tid: usize },
/// Get worker parameter
#[structopt(name = "get", version = garage_version())]
Get {
/// Gather variable values from all nodes
#[structopt(short = "a", long = "all-nodes")]
all_nodes: bool,
/// Variable name to get, or none to get all variables
variable: Option<String>,
},
/// Set worker parameter
#[structopt(name = "set", version = garage_version())]
Set {
#[structopt(subcommand)]
opt: WorkerSetCmd,
/// Set variable values on all nodes
#[structopt(short = "a", long = "all-nodes")]
all_nodes: bool,
/// Variable node to set
variable: String,
/// Value to set the variable to
value: String,
},
}
@@ -535,14 +550,33 @@ pub struct WorkerListOpt {
}
#[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 },
pub enum BlockOperation {
/// List all blocks that currently have a resync error
#[structopt(name = "list-errors", version = garage_version())]
ListErrors,
/// Get detailed information about a single block
#[structopt(name = "info", version = garage_version())]
Info {
/// Hash of the block for which to retrieve information
hash: String,
},
/// Retry now the resync of one or many blocks
#[structopt(name = "retry-now", version = garage_version())]
RetryNow {
/// Retry all blocks that have a resync error
#[structopt(long = "all")]
all: bool,
/// Hashes of the block to retry to resync now
blocks: Vec<String>,
},
/// Delete all objects referencing a missing block
#[structopt(name = "purge", version = garage_version())]
Purge {
/// Mandatory to confirm this operation
#[structopt(long = "yes")]
yes: bool,
/// Hashes of the block to purge
#[structopt(required = true)]
blocks: Vec<String>,
},
}
+148 -23
View File
@@ -3,14 +3,17 @@ use std::time::Duration;
use garage_util::background::*;
use garage_util::crdt::*;
use garage_util::data::Uuid;
use garage_util::data::*;
use garage_util::error::*;
use garage_util::formater::format_table;
use garage_util::time::*;
use garage_block::manager::BlockResyncErrorInfo;
use garage_model::bucket_table::*;
use garage_model::key_table::*;
use garage_model::s3::object_table::{BYTES, OBJECTS, UNFINISHED_UPLOADS};
use garage_model::s3::version_table::Version;
use crate::cli::structs::WorkerListOpt;
@@ -241,7 +244,7 @@ pub fn find_matching_node(
}
}
pub fn print_worker_info(wi: HashMap<usize, WorkerInfo>, wlo: WorkerListOpt) {
pub fn print_worker_list(wi: HashMap<usize, WorkerInfo>, wlo: WorkerListOpt) {
let mut wi = wi.into_iter().collect::<Vec<_>>();
wi.sort_by_key(|(tid, info)| {
(
@@ -254,7 +257,7 @@ pub fn print_worker_info(wi: HashMap<usize, WorkerInfo>, wlo: WorkerListOpt) {
)
});
let mut table = vec![];
let mut table = vec!["TID\tState\tName\tTranq\tDone\tQueue\tErrors\tConsec\tLast".to_string()];
for (tid, info) in wi.iter() {
if wlo.busy && !matches!(info.state, WorkerState::Busy | WorkerState::Throttled(_)) {
continue;
@@ -263,33 +266,155 @@ pub fn print_worker_info(wi: HashMap<usize, WorkerInfo>, wlo: WorkerListOpt) {
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
let err_ago = 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 {
.map(|(_, t)| tf.convert(Duration::from_millis(now_msec() - t)))
.unwrap_or_default();
let (total_err, consec_err) = if info.errors > 0 {
(info.errors.to_string(), info.consecutive_errors.to_string())
} else {
("-".into(), "-".into())
};
table.push(format!(
"{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}",
tid,
info.state,
info.name,
info.status
.tranquility
.as_ref()
.map(ToString::to_string)
.unwrap_or_else(|| "-".into()),
info.status.progress.as_deref().unwrap_or("-"),
info.status
.queue_length
.as_ref()
.map(ToString::to_string)
.unwrap_or_else(|| "-".into()),
total_err,
consec_err,
err_ago,
));
}
format_table(table);
}
pub fn print_worker_info(tid: usize, info: WorkerInfo) {
let mut table = vec![];
table.push(format!("Task id:\t{}", tid));
table.push(format!("Worker name:\t{}", info.name));
match info.state {
WorkerState::Throttled(t) => {
table.push(format!(
"\t\t {} consecutive errors ({} total), last {}",
info.consecutive_errors, info.errors, err_ago,
"Worker state:\tBusy (throttled, paused for {:.3}s)",
t
));
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));
}
s => {
table.push(format!("Worker state:\t{}", s));
}
};
if let Some(tql) = info.status.tranquility {
table.push(format!("Tranquility:\t{}", tql));
}
table.push("".into());
table.push(format!("Total errors:\t{}", info.errors));
table.push(format!("Consecutive errs:\t{}", info.consecutive_errors));
if let Some((s, t)) = info.last_error {
table.push(format!("Last error:\t{}", s));
let tf = timeago::Formatter::new();
table.push(format!(
"Last error time:\t{}",
tf.convert(Duration::from_millis(now_msec() - t))
));
}
table.push("".into());
if let Some(p) = info.status.progress {
table.push(format!("Progress:\t{}", p));
}
if let Some(ql) = info.status.queue_length {
table.push(format!("Queue length:\t{}", ql));
}
if let Some(pe) = info.status.persistent_errors {
table.push(format!("Persistent errors:\t{}", pe));
}
for (i, s) in info.status.freeform.iter().enumerate() {
if i == 0 {
if table.last() != Some(&"".into()) {
table.push("".into());
}
table.push(format!("Message:\t{}", s));
} else {
table.push(format!("\t{}", s));
}
}
format_table(table);
}
pub fn print_worker_vars(wv: Vec<(Uuid, String, String)>) {
let table = wv
.into_iter()
.map(|(n, k, v)| format!("{:?}\t{}\t{}", n, k, v))
.collect::<Vec<_>>();
format_table(table);
}
pub fn print_block_error_list(el: Vec<BlockResyncErrorInfo>) {
let now = now_msec();
let tf = timeago::Formatter::new();
let mut tf2 = timeago::Formatter::new();
tf2.ago("");
let mut table = vec!["Hash\tRC\tErrors\tLast error\tNext try".into()];
for e in el {
table.push(format!(
"{}\t{}\t{}\t{}\tin {}",
hex::encode(e.hash.as_slice()),
e.refcount,
e.error_count,
tf.convert(Duration::from_millis(now - e.last_try)),
tf2.convert(Duration::from_millis(e.next_try - now))
));
}
format_table(table);
}
pub fn print_block_info(hash: Hash, refcount: u64, versions: Vec<Result<Version, Uuid>>) {
println!("Block hash: {}", hex::encode(hash.as_slice()));
println!("Refcount: {}", refcount);
println!();
let mut table = vec!["Version\tBucket\tKey\tDeleted".into()];
let mut nondeleted_count = 0;
for v in versions.iter() {
match v {
Ok(ver) => {
table.push(format!(
"{:?}\t{:?}\t{}\t{:?}",
ver.uuid,
ver.bucket_id,
ver.key,
ver.deleted.get()
));
if !ver.deleted.get() {
nondeleted_count += 1;
}
}
Err(vh) => {
table.push(format!("{:?}\t\t\tyes", vh));
}
}
}
format_table(table);
if refcount != nondeleted_count {
println!();
println!("Warning: refcount does not match number of non-deleted versions");
}
}
+21 -15
View File
@@ -127,9 +127,16 @@ async fn main() {
std::process::abort();
}));
// Parse arguments and dispatch command line
let opt = Opt::from_clap(&Opt::clap().version(version.as_str()).get_matches());
// Initialize logging as well as other libraries used in Garage
if std::env::var("RUST_LOG").is_err() {
std::env::set_var("RUST_LOG", "netapp=info,garage=info")
let default_log = match &opt.cmd {
Command::Server => "netapp=info,garage=info",
_ => "netapp=warn,garage=warn",
};
std::env::set_var("RUST_LOG", default_log)
}
tracing_subscriber::fmt()
.with_writer(std::io::stderr)
@@ -137,9 +144,6 @@ async fn main() {
.init();
sodiumoxide::init().expect("Unable to init sodiumoxide");
// Parse arguments and dispatch command line
let opt = Opt::from_clap(&Opt::clap().version(version.as_str()).get_matches());
let res = match opt.cmd {
Command::Server => server::run_server(opt.config_file).await,
Command::OfflineRepair(repair_opt) => {
@@ -182,9 +186,9 @@ async fn cli_command(opt: Opt) -> Result<(), Error> {
let netapp = NetApp::new(GARAGE_VERSION_TAG, network_key, sk);
// Find and parse the address of the target host
let (id, addr) = if let Some(h) = opt.rpc_host {
let (id, addr, is_default_addr) = if let Some(h) = opt.rpc_host {
let (id, addrs) = parse_and_resolve_peer_addr(&h).ok_or_else(|| format!("Invalid RPC remote node identifier: {}. Expected format is <pubkey>@<IP or hostname>:<port>.", h))?;
(id, addrs[0])
(id, addrs[0], false)
} else {
let node_id = garage_rpc::system::read_node_id(&config.as_ref().unwrap().metadata_dir)
.err_context(READ_KEY_ERROR)?;
@@ -195,24 +199,26 @@ async fn cli_command(opt: Opt) -> Result<(), Error> {
.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)
(node_id, a, false)
} else {
let default_addr = SocketAddr::new(
"127.0.0.1".parse().unwrap(),
config.as_ref().unwrap().rpc_bind_addr.port(),
);
warn!(
"Trying to contact Garage node at default address {}",
default_addr
);
warn!("If this doesn't work, consider adding rpc_public_addr in your config file or specifying the -h command line parameter.");
(node_id, default_addr)
(node_id, default_addr, true)
}
};
// Connect to target host
netapp.clone().try_connect(addr, id).await
.err_context("Unable to connect to destination RPC host. Check that you are using the same value of rpc_secret as them, and that you have their correct public key.")?;
if let Err(e) = netapp.clone().try_connect(addr, id).await {
if is_default_addr {
warn!(
"Tried to contact Garage node at default address {}, which didn't work. If that address is wrong, consider setting rpc_public_addr in your config file.",
addr
);
}
Err(e).err_context("Unable to connect to destination RPC host. Check that you are using the same value of rpc_secret as them, and that you have their correct public key.")?;
}
let system_rpc_endpoint = netapp.endpoint::<SystemRpc, ()>(SYSTEM_RPC_PATH.into());
let admin_rpc_endpoint = netapp.endpoint::<AdminRpc, ()>(ADMIN_RPC_PATH.into());
+2 -15
View File
@@ -1,8 +1,5 @@
use std::path::PathBuf;
use tokio::sync::watch;
use garage_util::background::*;
use garage_util::config::*;
use garage_util::error::*;
@@ -20,12 +17,8 @@ pub async fn offline_repair(config_file: PathBuf, opt: OfflineRepairOpt) -> Resu
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)?;
let garage = Garage::new(config)?;
info!("Launching repair operation...");
match opt.what {
@@ -43,13 +36,7 @@ pub async fn offline_repair(config_file: PathBuf, opt: OfflineRepairOpt) -> Resu
}
}
info!("Repair operation finished, shutting down Garage internals...");
done_tx.send(true).unwrap();
drop(garage);
await_background_done.await?;
info!("Cleaning up...");
info!("Repair operation finished, shutting down...");
Ok(())
}
+53 -46
View File
@@ -12,38 +12,37 @@ use garage_model::s3::version_table::*;
use garage_table::*;
use garage_util::background::*;
use garage_util::error::Error;
use garage_util::migrate::Migrate;
use crate::*;
pub async fn launch_online_repair(garage: Arc<Garage>, opt: RepairOpt) {
pub async fn launch_online_repair(
garage: &Arc<Garage>,
bg: &BackgroundRunner,
opt: RepairOpt,
) -> Result<(), Error> {
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();
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()));
bg.spawn_worker(RepairVersionsWorker::new(garage.clone()));
}
RepairWhat::BlockRefs => {
info!("Repairing the block refs table");
garage
.background
.spawn_worker(RepairBlockrefsWorker::new(garage.clone()));
bg.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(),
));
bg.spawn_worker(garage_block::repair::RepairWorker::new(
garage.block_manager.clone(),
));
}
RepairWhat::Scrub { cmd } => {
let cmd = match cmd {
@@ -52,13 +51,18 @@ pub async fn launch_online_repair(garage: Arc<Garage>, opt: RepairOpt) {
ScrubCmd::Resume => ScrubWorkerCommand::Resume,
ScrubCmd::Cancel => ScrubWorkerCommand::Cancel,
ScrubCmd::SetTranquility { tranquility } => {
ScrubWorkerCommand::SetTranquility(tranquility)
garage
.block_manager
.scrub_persister
.set_with(|x| x.tranquility = tranquility)?;
return Ok(());
}
};
info!("Sending command to scrub worker: {:?}", cmd);
garage.block_manager.send_scrub_command(cmd).await;
garage.block_manager.send_scrub_command(cmd).await?;
}
}
Ok(())
}
// ----
@@ -85,25 +89,23 @@ impl Worker for RepairVersionsWorker {
"Version repair worker".into()
}
fn info(&self) -> Option<String> {
Some(format!("{} items done", self.counter))
fn status(&self) -> WorkerStatus {
WorkerStatus {
progress: Some(self.counter.to_string()),
..Default::default()
}
}
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
}
let (item_bytes, next_pos) = match self.garage.version_table.data.store.get_gt(&self.pos)? {
Some((k, v)) => (v, k),
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)?;
let version = Version::decode(&item_bytes).ok_or_message("Cannot decode Version")?;
if !version.deleted.get() {
let object = self
.garage
@@ -131,10 +133,13 @@ impl Worker for RepairVersionsWorker {
}
}
self.counter += 1;
self.pos = next_pos;
Ok(WorkerState::Busy)
}
async fn wait_for_work(&mut self, _must_exit: &watch::Receiver<bool>) -> WorkerState {
async fn wait_for_work(&mut self) -> WorkerState {
unreachable!()
}
}
@@ -163,25 +168,24 @@ impl Worker for RepairBlockrefsWorker {
"Block refs repair worker".into()
}
fn info(&self) -> Option<String> {
Some(format!("{} items done", self.counter))
fn status(&self) -> WorkerStatus {
WorkerStatus {
progress: Some(self.counter.to_string()),
..Default::default()
}
}
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);
}
};
let (item_bytes, next_pos) =
match self.garage.block_ref_table.data.store.get_gt(&self.pos)? {
Some((k, v)) => (v, k),
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)?;
let block_ref = BlockRef::decode(&item_bytes).ok_or_message("Cannot decode BlockRef")?;
if !block_ref.deleted.get() {
let version = self
.garage
@@ -206,10 +210,13 @@ impl Worker for RepairBlockrefsWorker {
}
}
self.counter += 1;
self.pos = next_pos;
Ok(WorkerState::Busy)
}
async fn wait_for_work(&mut self, _must_exit: &watch::Receiver<bool>) -> WorkerState {
async fn wait_for_work(&mut self) -> WorkerState {
unreachable!()
}
}
+7 -4
View File
@@ -35,12 +35,15 @@ pub async fn run_server(config_file: PathBuf) -> Result<(), Error> {
#[cfg(feature = "metrics")]
let metrics_exporter = opentelemetry_prometheus::exporter().init();
info!("Initializing Garage main data store...");
let garage = Garage::new(config.clone())?;
info!("Initializing background runner...");
let watch_cancel = watch_shutdown_signal();
let (background, await_background_done) = BackgroundRunner::new(16, watch_cancel.clone());
let (background, await_background_done) = BackgroundRunner::new(watch_cancel.clone());
info!("Initializing Garage main data store...");
let garage = Garage::new(config.clone(), background)?;
info!("Spawning Garage workers...");
garage.spawn_workers(&background);
if config.admin.trace_sink.is_some() {
info!("Initialize tracing...");
@@ -63,7 +66,7 @@ pub async fn run_server(config_file: PathBuf) -> Result<(), Error> {
let run_system = tokio::spawn(garage.system.clone().run(watch_cancel.clone()));
info!("Create admin RPC handler...");
AdminRpcHandler::new(garage.clone());
AdminRpcHandler::new(garage.clone(), background.clone());
// ---- Launch public-facing API servers ----
+22
View File
@@ -1,4 +1,5 @@
use crate::common;
use crate::common::ext::CommandExt;
use aws_sdk_s3::model::BucketLocationConstraint;
use aws_sdk_s3::output::DeleteBucketOutput;
@@ -8,6 +9,27 @@ async fn test_bucket_all() {
let bucket_name = "hello";
{
// Check bucket cannot be created if not authorized
ctx.garage
.command()
.args(["key", "deny"])
.args(["--create-bucket", &ctx.garage.key.id])
.quiet()
.expect_success_output("Could not deny key to create buckets");
// Try create bucket, should fail
let r = ctx.client.create_bucket().bucket(bucket_name).send().await;
assert!(r.is_err());
}
{
// Now allow key to create bucket
ctx.garage
.command()
.args(["key", "allow"])
.args(["--create-bucket", &ctx.garage.key.id])
.quiet()
.expect_success_output("Could not deny key to create buckets");
// Create bucket
//@TODO check with an invalid bucket name + with an already existing bucket
let r = ctx
@@ -1,6 +1,7 @@
use std::collections::HashMap;
use crate::common;
use crate::common::ext::CommandExt;
use common::custom_requester::BodySignature;
use hyper::Method;
@@ -105,6 +106,13 @@ async fn test_create_bucket_streaming() {
let ctx = common::context();
let bucket = "createbucket-streaming";
ctx.garage
.command()
.args(["key", "allow"])
.args(["--create-bucket", &ctx.garage.key.id])
.quiet()
.expect_success_output("Could not allow key to create buckets");
{
// create bucket
let _ = ctx
+1 -1
View File
@@ -22,7 +22,7 @@ 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 }
garage_util = { version = "0.8.1", path = "../util", optional = true }
[features]
+6 -7
View File
@@ -1,6 +1,6 @@
[package]
name = "garage_model"
version = "0.8.0"
version = "0.8.1"
authors = ["Alex Auvolat <alex@adnab.me>"]
edition = "2018"
license = "AGPL-3.0"
@@ -14,11 +14,11 @@ path = "lib.rs"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
garage_db = { version = "0.8.0", default-features = false, 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" }
garage_db = { version = "0.8.1", default-features = false, path = "../db" }
garage_rpc = { version = "0.8.1", path = "../rpc" }
garage_table = { version = "0.8.1", path = "../table" }
garage_block = { version = "0.8.1", path = "../block" }
garage_util = { version = "0.8.1", path = "../util" }
async-trait = "0.1.7"
arc-swap = "1.0"
@@ -30,7 +30,6 @@ tracing = "0.1.30"
rand = "0.8"
zstd = { version = "0.9", default-features = false }
rmp-serde = "0.15"
serde = { version = "1.0", default-features = false, features = ["derive", "rc"] }
serde_bytes = "0.11"
+16 -8
View File
@@ -1,18 +1,26 @@
use serde::{Deserialize, Serialize};
use garage_util::data::*;
use garage_table::crdt::*;
use garage_table::*;
/// The bucket alias table holds the names given to buckets
/// in the global namespace.
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct BucketAlias {
name: String,
pub state: crdt::Lww<Option<Uuid>>,
mod v08 {
use garage_util::crdt;
use garage_util::data::Uuid;
use serde::{Deserialize, Serialize};
/// The bucket alias table holds the names given to buckets
/// in the global namespace.
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct BucketAlias {
pub(super) name: String,
pub state: crdt::Lww<Option<Uuid>>,
}
impl garage_util::migrate::InitialFormat for BucketAlias {}
}
pub use v08::*;
impl BucketAlias {
pub fn new(name: String, ts: u64, bucket_id: Option<Uuid>) -> Option<Self> {
if !is_valid_bucket_name(&name) {
+74 -65
View File
@@ -1,5 +1,3 @@
use serde::{Deserialize, Serialize};
use garage_table::crdt::*;
use garage_table::*;
use garage_util::data::*;
@@ -7,71 +5,82 @@ use garage_util::time::*;
use crate::permission::BucketKeyPerm;
/// A bucket is a collection of objects
///
/// 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, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Bucket {
/// ID of the bucket
pub id: Uuid,
/// State, and configuration if not deleted, of the bucket
pub state: crdt::Deletable<BucketParams>,
mod v08 {
use crate::permission::BucketKeyPerm;
use garage_util::crdt;
use garage_util::data::Uuid;
use serde::{Deserialize, Serialize};
/// A bucket is a collection of objects
///
/// 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, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Bucket {
/// ID of the bucket
pub id: Uuid,
/// State, and configuration if not deleted, of the bucket
pub state: crdt::Deletable<BucketParams>,
}
/// Configuration for a bucket
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct BucketParams {
/// Bucket's creation date
pub creation_date: u64,
/// Map of key with access to the bucket, and what kind of access they give
pub authorized_keys: crdt::Map<String, BucketKeyPerm>,
/// Map of aliases that are or have been given to this bucket
/// in the global namespace
/// (not authoritative: this is just used as an indication to
/// map back to aliases when doing ListBuckets)
pub aliases: crdt::LwwMap<String, bool>,
/// Map of aliases that are or have been given to this bucket
/// in namespaces local to keys
/// key = (access key id, alias name)
pub local_aliases: crdt::LwwMap<(String, String), bool>,
/// Whether this bucket is allowed for website access
/// (under all of its global alias names),
/// and if so, the website configuration XML document
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)]
pub struct WebsiteConfig {
pub index_document: String,
pub error_document: Option<String>,
}
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct CorsRule {
pub id: Option<String>,
pub max_age_seconds: Option<u64>,
pub allow_origins: Vec<String>,
pub allow_methods: Vec<String>,
pub allow_headers: Vec<String>,
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 garage_util::migrate::InitialFormat for Bucket {}
}
/// Configuration for a bucket
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct BucketParams {
/// Bucket's creation date
pub creation_date: u64,
/// Map of key with access to the bucket, and what kind of access they give
pub authorized_keys: crdt::Map<String, BucketKeyPerm>,
/// Map of aliases that are or have been given to this bucket
/// in the global namespace
/// (not authoritative: this is just used as an indication to
/// map back to aliases when doing ListBuckets)
pub aliases: crdt::LwwMap<String, bool>,
/// Map of aliases that are or have been given to this bucket
/// in namespaces local to keys
/// key = (access key id, alias name)
pub local_aliases: crdt::LwwMap<(String, String), bool>,
/// Whether this bucket is allowed for website access
/// (under all of its global alias names),
/// and if so, the website configuration XML document
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)]
pub struct WebsiteConfig {
pub index_document: String,
pub error_document: Option<String>,
}
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct CorsRule {
pub id: Option<String>,
pub max_age_seconds: Option<u64>,
pub allow_origins: Vec<String>,
pub allow_methods: Vec<String>,
pub allow_headers: Vec<String>,
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>,
}
pub use v08::*;
impl AutoCrdt for BucketQuotas {
const WARN_IF_DIFFERENT: bool = true;
+59 -15
View File
@@ -8,10 +8,10 @@ use garage_util::background::*;
use garage_util::config::*;
use garage_util::error::*;
use garage_rpc::replication_mode::ReplicationMode;
use garage_rpc::system::System;
use garage_block::manager::*;
use garage_table::replication::ReplicationMode;
use garage_table::replication::TableFullReplication;
use garage_table::replication::TableShardedReplication;
use garage_table::*;
@@ -27,17 +27,20 @@ use crate::index_counter::*;
use crate::key_table::*;
#[cfg(feature = "k2v")]
use crate::k2v::{item_table::*, poll::*, rpc::*};
use crate::k2v::{history_table::*, item_table::*, sub::*, rpc::*};
/// An entire Garage full of data
pub struct Garage {
/// The parsed configuration Garage is running
pub config: Config,
/// The set of background variables that can be viewed/modified at runtime
pub bg_vars: vars::BgVars,
/// The replication mode of this cluster
pub replication_mode: ReplicationMode,
/// The local database
pub db: db::Db,
/// A background job runner
pub background: Arc<BackgroundRunner>,
/// The membership manager
pub system: Arc<System>,
/// The block manager
@@ -67,6 +70,8 @@ pub struct Garage {
pub struct GarageK2V {
/// Table containing K2V items
pub item_table: Arc<Table<K2VItemTable, TableShardedReplication>>,
/// Table containing K2V modification history
pub history_table: Arc<Table<K2VHistoryTable, TableShardedReplication>>,
/// Indexing table containing K2V item counters
pub counter_table: Arc<IndexCounter<K2VItem>>,
/// K2V RPC handler
@@ -75,7 +80,7 @@ pub struct GarageK2V {
impl Garage {
/// Create and run garage
pub fn new(config: Config, background: Arc<BackgroundRunner>) -> Result<Arc<Self>, Error> {
pub fn new(config: Config) -> 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")?;
@@ -164,12 +169,7 @@ impl Garage {
.expect("Invalid replication_mode in config file.");
info!("Initialize membership management system...");
let system = System::new(
network_key,
background.clone(),
replication_mode.replication_factor(),
&config,
)?;
let system = System::new(network_key, replication_mode, &config)?;
let data_rep_param = TableShardedReplication {
system: system.clone(),
@@ -227,7 +227,6 @@ impl Garage {
info!("Initialize version_table...");
let version_table = Table::new(
VersionTable {
background: background.clone(),
block_ref_table: block_ref_table.clone(),
},
meta_rep_param.clone(),
@@ -242,7 +241,6 @@ impl Garage {
#[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(),
},
@@ -255,11 +253,16 @@ impl Garage {
#[cfg(feature = "k2v")]
let k2v = GarageK2V::new(system.clone(), &db, meta_rep_param);
// Initialize bg vars
let mut bg_vars = vars::BgVars::new();
block_manager.register_bg_vars(&mut bg_vars);
// -- done --
Ok(Arc::new(Self {
config,
bg_vars,
replication_mode,
db,
background,
system,
block_manager,
bucket_table,
@@ -274,6 +277,22 @@ impl Garage {
}))
}
pub fn spawn_workers(&self, bg: &BackgroundRunner) {
self.block_manager.spawn_workers(bg);
self.bucket_table.spawn_workers(bg);
self.bucket_alias_table.spawn_workers(bg);
self.key_table.spawn_workers(bg);
self.object_table.spawn_workers(bg);
self.object_counter_table.spawn_workers(bg);
self.version_table.spawn_workers(bg);
self.block_ref_table.spawn_workers(bg);
#[cfg(feature = "k2v")]
self.k2v.spawn_workers(bg);
}
pub fn bucket_helper(&self) -> helper::bucket::BucketHelper {
helper::bucket::BucketHelper(self)
}
@@ -288,24 +307,49 @@ 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.clone(),
system.clone(),
db,
);
info!("Initialize K2V history table...");
let history_table = Table::new(
K2VHistoryTable {
subscriptions: subscriptions.clone(),
},
meta_rep_param,
system.clone(),
db,
);
let rpc = K2VRpcHandler::new(system, item_table.clone(), subscriptions);
info!("Initialize K2V RPC handler...");
let rpc = K2VRpcHandler::new(
system,
db,
item_table.clone(),
history_table.clone(),
subscriptions,
);
Self {
item_table,
history_table,
counter_table,
rpc,
}
}
pub fn spawn_workers(&self, bg: &BackgroundRunner) {
self.item_table.spawn_workers(bg);
self.counter_table.spawn_workers(bg);
}
}
+64 -168
View File
@@ -1,19 +1,18 @@
use core::ops::Bound;
use std::collections::{hash_map, BTreeMap, HashMap};
use std::collections::{BTreeMap, HashMap};
use std::marker::PhantomData;
use std::sync::Arc;
use async_trait::async_trait;
use serde::{Deserialize, Serialize};
use tokio::sync::{mpsc, watch};
use garage_db as db;
use garage_rpc::ring::Ring;
use garage_rpc::system::System;
use garage_util::background::*;
use garage_util::background::BackgroundRunner;
use garage_util::data::*;
use garage_util::error::*;
use garage_util::migrate::Migrate;
use garage_util::time::*;
use garage_table::crdt::*;
@@ -31,14 +30,44 @@ pub trait CountedItem: Clone + PartialEq + Send + Sync + 'static {
fn counts(&self) -> Vec<(&'static str, i64)>;
}
/// A counter entry in the global table
#[derive(Clone, PartialEq, Debug, Serialize, Deserialize)]
pub struct CounterEntry<T: CountedItem> {
pub pk: T::CP,
pub sk: T::CS,
pub values: BTreeMap<String, CounterValue>,
mod v08 {
use super::CountedItem;
use garage_util::data::Uuid;
use serde::{Deserialize, Serialize};
use std::collections::BTreeMap;
// ---- Global part (the table everyone queries) ----
/// A counter entry in the global table
#[derive(Clone, PartialEq, Debug, Serialize, Deserialize)]
pub struct CounterEntry<T: CountedItem> {
pub pk: T::CP,
pub sk: T::CS,
pub values: BTreeMap<String, CounterValue>,
}
/// A counter entry in the global table
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct CounterValue {
pub node_values: BTreeMap<Uuid, (u64, i64)>,
}
impl<T: CountedItem> garage_util::migrate::InitialFormat for CounterEntry<T> {}
// ---- Local part (the counter we maintain transactionnaly on each node) ----
#[derive(PartialEq, Clone, Debug, Serialize, Deserialize)]
pub(super) struct LocalCounterEntry<T: CountedItem> {
pub(super) pk: T::CP,
pub(super) sk: T::CS,
pub(super) values: BTreeMap<String, (u64, i64)>,
}
impl<T: CountedItem> garage_util::migrate::InitialFormat for LocalCounterEntry<T> {}
}
pub use v08::*;
impl<T: CountedItem> Entry<T::CP, T::CS> for CounterEntry<T> {
fn partition_key(&self) -> &T::CP {
&self.pk
@@ -80,12 +109,6 @@ impl<T: CountedItem> CounterEntry<T> {
}
}
/// A counter entry in the global table
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct CounterValue {
pub node_values: BTreeMap<Uuid, (u64, i64)>,
}
impl<T: CountedItem> Crdt for CounterEntry<T> {
fn merge(&mut self, other: &Self) {
for (name, e2) in other.values.iter() {
@@ -142,7 +165,6 @@ impl<T: CountedItem> TableSchema for CounterTable<T> {
pub struct IndexCounter<T: CountedItem> {
this_node: Uuid,
local_counter: db::Tree,
propagate_tx: mpsc::UnboundedSender<(T::CP, T::CS, LocalCounterEntry<T>)>,
pub table: Arc<Table<CounterTable<T>, TableShardedReplication>>,
}
@@ -152,16 +174,11 @@ impl<T: CountedItem> IndexCounter<T> {
replication: TableShardedReplication,
db: &db::Db,
) -> Arc<Self> {
let background = system.background.clone();
let (propagate_tx, propagate_rx) = mpsc::unbounded_channel();
let this = Arc::new(Self {
Arc::new(Self {
this_node: system.id,
local_counter: db
.open_tree(format!("local_counter_v2:{}", T::COUNTER_TABLE_NAME))
.expect("Unable to open local counter tree"),
propagate_tx,
table: Table::new(
CounterTable {
_phantom_t: Default::default(),
@@ -170,16 +187,11 @@ impl<T: CountedItem> IndexCounter<T> {
system,
db,
),
});
})
}
background.spawn_worker(IndexPropagatorWorker {
index_counter: this.clone(),
propagate_rx,
buf: HashMap::new(),
errors: 0,
});
this
pub fn spawn_workers(&self, bg: &BackgroundRunner) {
self.table.spawn_workers(bg);
}
pub fn count(
@@ -208,11 +220,9 @@ impl<T: CountedItem> IndexCounter<T> {
let tree_key = self.table.data.tree_key(pk, sk);
let mut entry = match tx.get(&self.local_counter, &tree_key[..])? {
Some(old_bytes) => {
rmp_serde::decode::from_read_ref::<_, LocalCounterEntry<T>>(&old_bytes)
.map_err(Error::RmpDecode)
.map_err(db::TxError::Abort)?
}
Some(old_bytes) => LocalCounterEntry::<T>::decode(&old_bytes)
.ok_or_message("Cannot decode local counter entry")
.map_err(db::TxError::Abort)?,
None => LocalCounterEntry {
pk: pk.clone(),
sk: sk.clone(),
@@ -227,17 +237,14 @@ impl<T: CountedItem> IndexCounter<T> {
ent.1 += *inc;
}
let new_entry_bytes = rmp_to_vec_all_named(&entry)
let new_entry_bytes = entry
.encode()
.map_err(Error::RmpEncode)
.map_err(db::TxError::Abort)?;
tx.insert(&self.local_counter, &tree_key[..], new_entry_bytes)?;
if let Err(e) = self.propagate_tx.send((pk.clone(), sk.clone(), entry)) {
error!(
"Could not propagate updated counter values, failed to send to channel: {}",
e
);
}
let dist_entry = entry.into_counter_entry(self.this_node);
self.table.queue_insert(tx, &dist_entry)?;
Ok(())
}
@@ -250,23 +257,6 @@ impl<T: CountedItem> IndexCounter<T> {
TS: TableSchema<E = T>,
TR: TableReplication,
{
let save_counter_entry = |entry: CounterEntry<T>| -> Result<(), Error> {
let entry_k = self
.table
.data
.tree_key(entry.partition_key(), entry.sort_key());
self.table
.data
.update_entry_with(&entry_k, |ent| match ent {
Some(mut ent) => {
ent.merge(&entry);
ent
}
None => entry.clone(),
})?;
Ok(())
};
// 1. Set all old local counters to zero
let now = now_msec();
let mut next_start: Option<Vec<u8>> = None;
@@ -289,20 +279,22 @@ impl<T: CountedItem> IndexCounter<T> {
info!("zeroing old counters... ({})", hex::encode(&batch[0].0));
for (local_counter_k, local_counter) in batch {
let mut local_counter =
rmp_serde::decode::from_read_ref::<_, LocalCounterEntry<T>>(&local_counter)?;
let mut local_counter = LocalCounterEntry::<T>::decode(&local_counter)
.ok_or_message("Cannot decode local counter entry")?;
for (_, tv) in local_counter.values.iter_mut() {
tv.0 = std::cmp::max(tv.0 + 1, now);
tv.1 = 0;
}
let local_counter_bytes = rmp_to_vec_all_named(&local_counter)?;
let local_counter_bytes = local_counter.encode()?;
self.local_counter
.insert(&local_counter_k, &local_counter_bytes)?;
let counter_entry = local_counter.into_counter_entry(self.this_node);
save_counter_entry(counter_entry)?;
self.local_counter
.db()
.transaction(|mut tx| self.table.queue_insert(&mut tx, &counter_entry))?;
next_start = Some(local_counter_k);
}
@@ -343,9 +335,8 @@ impl<T: CountedItem> IndexCounter<T> {
let local_counter_key = self.table.data.tree_key(pk, sk);
let mut local_counter = match self.local_counter.get(&local_counter_key)? {
Some(old_bytes) => {
let ent = rmp_serde::decode::from_read_ref::<_, LocalCounterEntry<T>>(
&old_bytes,
)?;
let ent = LocalCounterEntry::<T>::decode(&old_bytes)
.ok_or_message("Cannot decode local counter entry")?;
assert!(ent.pk == *pk);
assert!(ent.sk == *sk);
ent
@@ -362,12 +353,14 @@ impl<T: CountedItem> IndexCounter<T> {
tv.1 += v;
}
let local_counter_bytes = rmp_to_vec_all_named(&local_counter)?;
let local_counter_bytes = local_counter.encode()?;
self.local_counter
.insert(&local_counter_key, local_counter_bytes)?;
let counter_entry = local_counter.into_counter_entry(self.this_node);
save_counter_entry(counter_entry)?;
self.local_counter
.db()
.transaction(|mut tx| self.table.queue_insert(&mut tx, &counter_entry))?;
next_start = Some(counted_entry_k);
}
@@ -378,104 +371,7 @@ impl<T: CountedItem> IndexCounter<T> {
}
}
struct IndexPropagatorWorker<T: CountedItem> {
index_counter: Arc<IndexCounter<T>>,
propagate_rx: mpsc::UnboundedReceiver<(T::CP, T::CS, LocalCounterEntry<T>)>,
buf: HashMap<Vec<u8>, CounterEntry<T>>,
errors: usize,
}
impl<T: CountedItem> IndexPropagatorWorker<T> {
fn add_ent(&mut self, pk: T::CP, sk: T::CS, counters: LocalCounterEntry<T>) {
let tree_key = self.index_counter.table.data.tree_key(&pk, &sk);
let dist_entry = counters.into_counter_entry(self.index_counter.this_node);
match self.buf.entry(tree_key) {
hash_map::Entry::Vacant(e) => {
e.insert(dist_entry);
}
hash_map::Entry::Occupied(mut e) => {
e.get_mut().merge(&dist_entry);
}
}
}
}
#[async_trait]
impl<T: CountedItem> Worker for IndexPropagatorWorker<T> {
fn name(&self) -> String {
format!("{} index counter propagator", T::COUNTER_TABLE_NAME)
}
fn info(&self) -> Option<String> {
if !self.buf.is_empty() {
Some(format!("{} items in queue", self.buf.len()))
} else {
None
}
}
async fn work(&mut self, must_exit: &mut watch::Receiver<bool>) -> Result<WorkerState, Error> {
// This loop batches updates to counters to be sent all at once.
// They are sent once the propagate_rx channel has been emptied (or is closed).
let closed = loop {
match self.propagate_rx.try_recv() {
Ok((pk, sk, counters)) => {
self.add_ent(pk, sk, counters);
}
Err(mpsc::error::TryRecvError::Empty) => break false,
Err(mpsc::error::TryRecvError::Disconnected) => break true,
}
};
if !self.buf.is_empty() {
let entries_k = self.buf.keys().take(100).cloned().collect::<Vec<_>>();
let entries = entries_k.iter().map(|k| self.buf.get(k).unwrap());
if let Err(e) = self.index_counter.table.insert_many(entries).await {
self.errors += 1;
if self.errors >= 2 && *must_exit.borrow() {
error!("({}) Could not propagate {} counter values: {}, these counters will not be updated correctly.", T::COUNTER_TABLE_NAME, self.buf.len(), e);
return Ok(WorkerState::Done);
}
// Propagate error up to worker manager, it will log it, increment a counter,
// and sleep for a certain delay (with exponential backoff), waiting for
// things to go back to normal
return Err(e);
} else {
for k in entries_k {
self.buf.remove(&k);
}
self.errors = 0;
}
return Ok(WorkerState::Busy);
} else if closed {
return Ok(WorkerState::Done);
} else {
return Ok(WorkerState::Idle);
}
}
async fn wait_for_work(&mut self, _must_exit: &watch::Receiver<bool>) -> WorkerState {
match self.propagate_rx.recv().await {
Some((pk, sk, counters)) => {
self.add_ent(pk, sk, counters);
WorkerState::Busy
}
None => match self.buf.is_empty() {
false => WorkerState::Busy,
true => WorkerState::Done,
},
}
}
}
#[derive(PartialEq, Clone, Debug, Serialize, Deserialize)]
struct LocalCounterEntry<T: CountedItem> {
pk: T::CP,
sk: T::CS,
values: BTreeMap<String, (u64, i64)>,
}
// ----
impl<T: CountedItem> LocalCounterEntry<T> {
fn into_counter_entry(self, this_node: Uuid) -> CounterEntry<T> {
+107
View File
@@ -0,0 +1,107 @@
use std::sync::Arc;
use garage_db as db;
use garage_table::crdt::*;
use garage_table::*;
use crate::k2v::sub::*;
mod v08 {
use crate::k2v::causality::K2VNodeId;
pub use crate::k2v::item_table::v08::{DvvsValue, K2VItem, K2VItemPartition};
use garage_util::crdt;
use serde::{Deserialize, Serialize};
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct K2VHistoryEntry {
// Partition key: the partition key of ins_item
/// The inserted item
pub ins_item: K2VItem,
/// Sort key: the node ID and its local counter
pub node_counter: K2VHistorySortKey,
/// The value of the node's local counter before this entry was updated
pub prev_counter: u64,
/// The timesamp of the update (!= counter, counters are incremented
/// by one, timestamps are real clock timestamps)
pub timestamp: u64,
/// Mark this history entry for deletion
pub deleted: crdt::Bool,
}
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct K2VHistorySortKey {
pub node: K2VNodeId,
pub counter: u64,
}
impl garage_util::migrate::InitialFormat for K2VHistoryEntry {
const VERSION_MARKER: &'static [u8] = b"Gk2vhe08";
}
}
pub use v08::*;
impl Crdt for K2VHistoryEntry {
fn merge(&mut self, other: &Self) {
self.ins_item.merge(&other.ins_item);
self.deleted.merge(&other.deleted);
}
}
impl SortKey for K2VHistorySortKey {
type B<'a> = [u8; 16];
fn sort_key(&self) -> [u8; 16] {
let mut ret = [0u8; 16];
ret[0..8].copy_from_slice(&u64::to_be_bytes(self.node));
ret[8..16].copy_from_slice(&u64::to_be_bytes(self.counter));
ret
}
}
impl Entry<K2VItemPartition, K2VHistorySortKey> for K2VHistoryEntry {
fn partition_key(&self) -> &K2VItemPartition {
&self.ins_item.partition
}
fn sort_key(&self) -> &K2VHistorySortKey {
&self.node_counter
}
fn is_tombstone(&self) -> bool {
self.deleted.get()
}
}
pub struct K2VHistoryTable {
pub(crate) subscriptions: Arc<SubscriptionManager>,
}
impl TableSchema for K2VHistoryTable {
const TABLE_NAME: &'static str = "k2v_history";
type P = K2VItemPartition;
type S = K2VHistorySortKey;
type E = K2VHistoryEntry;
type Filter = DeletedFilter;
fn updated(
&self,
_tx: &mut db::Transaction,
_old: Option<&Self::E>,
new: Option<&Self::E>,
) -> db::TxOpResult<()> {
if let Some(new_ent) = new {
self.subscriptions.notify_range(new_ent);
}
Ok(())
}
fn matches_filter(entry: &Self::E, filter: &Self::Filter) -> bool {
filter.apply(entry.deleted.get())
}
}
+42 -27
View File
@@ -1,7 +1,8 @@
use serde::{Deserialize, Serialize};
use std::collections::BTreeMap;
use std::sync::Arc;
use serde::{Deserialize, Serialize};
use garage_db as db;
use garage_util::data::*;
@@ -10,38 +11,49 @@ use garage_table::*;
use crate::index_counter::*;
use crate::k2v::causality::*;
use crate::k2v::poll::*;
use crate::k2v::sub::*;
pub const ENTRIES: &str = "entries";
pub const CONFLICTS: &str = "conflicts";
pub const VALUES: &str = "values";
pub const BYTES: &str = "bytes";
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct K2VItem {
pub partition: K2VItemPartition,
pub sort_key: String,
pub(super) mod v08 {
use crate::k2v::causality::K2VNodeId;
use garage_util::data::Uuid;
use serde::{Deserialize, Serialize};
use std::collections::BTreeMap;
items: BTreeMap<K2VNodeId, DvvsEntry>,
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct K2VItem {
pub partition: K2VItemPartition,
pub sort_key: String,
pub(super) items: BTreeMap<K2VNodeId, DvvsEntry>,
}
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize, Hash)]
pub struct K2VItemPartition {
pub bucket_id: Uuid,
pub partition_key: String,
}
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct DvvsEntry {
pub(super) t_discard: u64,
pub(super) values: Vec<(u64, DvvsValue)>,
}
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub enum DvvsValue {
Value(#[serde(with = "serde_bytes")] Vec<u8>),
Deleted,
}
impl garage_util::migrate::InitialFormat for K2VItem {}
}
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize, Hash)]
pub struct K2VItemPartition {
pub bucket_id: Uuid,
pub partition_key: String,
}
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
struct DvvsEntry {
t_discard: u64,
values: Vec<(u64, DvvsValue)>,
}
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub enum DvvsValue {
Value(#[serde(with = "serde_bytes")] Vec<u8>),
Deleted,
}
pub use v08::*;
impl K2VItem {
/// Creates a new K2VItem when no previous entry existed in the db
@@ -61,7 +73,8 @@ impl K2VItem {
this_node: Uuid,
context: &Option<CausalContext>,
new_value: DvvsValue,
) {
node_counter: u64,
) -> u64 {
if let Some(context) = context {
for (node, t_discard) in context.vector_clock.iter() {
if let Some(e) = self.items.get_mut(node) {
@@ -86,7 +99,9 @@ impl K2VItem {
values: vec![],
});
let t_prev = e.max_time();
e.values.push((t_prev + 1, new_value));
let t_new = std::cmp::max(node_counter + 1, t_prev + 1);
e.values.push((t_new, new_value));
t_new
}
/// Extract the causality context of a K2V Item
@@ -225,7 +240,7 @@ impl TableSchema for K2VItemTable {
// 2. Notify
if let Some(new_ent) = new {
self.subscriptions.notify(new_ent);
self.subscriptions.notify_item(new_ent);
}
Ok(())
+3 -1
View File
@@ -1,6 +1,8 @@
pub mod causality;
pub mod history_table;
pub mod item_table;
pub mod poll;
pub(crate) mod sub;
pub mod rpc;
-50
View File
@@ -1,50 +0,0 @@
use std::collections::HashMap;
use std::sync::Mutex;
use serde::{Deserialize, Serialize};
use tokio::sync::broadcast;
use crate::k2v::item_table::*;
#[derive(Debug, Hash, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct PollKey {
pub partition: K2VItemPartition,
pub sort_key: String,
}
#[derive(Default)]
pub struct SubscriptionManager {
subscriptions: Mutex<HashMap<PollKey, broadcast::Sender<K2VItem>>>,
}
impl SubscriptionManager {
pub fn new() -> Self {
Self::default()
}
pub fn subscribe(&self, key: &PollKey) -> broadcast::Receiver<K2VItem> {
let mut subs = self.subscriptions.lock().unwrap();
if let Some(s) = subs.get(key) {
s.subscribe()
} else {
let (tx, rx) = broadcast::channel(8);
subs.insert(key.clone(), tx);
rx
}
}
pub fn notify(&self, item: &K2VItem) {
let key = PollKey {
partition: item.partition.clone(),
sort_key: item.sort_key.clone(),
};
let mut subs = self.subscriptions.lock().unwrap();
if let Some(s) = subs.get(&key) {
if s.send(item.clone()).is_err() {
// no more subscribers, remove channel from here
// (we will re-create it later if we need to subscribe again)
subs.remove(&key);
}
}
}
}
+54 -12
View File
@@ -6,6 +6,7 @@
//! mean the vector clock gets much larger than needed).
use std::collections::HashMap;
use std::convert::TryInto;
use std::sync::Arc;
use std::time::Duration;
@@ -15,9 +16,12 @@ use futures::StreamExt;
use serde::{Deserialize, Serialize};
use tokio::select;
use garage_db as db;
use garage_util::crdt::*;
use garage_util::data::*;
use garage_util::error::*;
use garage_util::time::*;
use garage_rpc::system::System;
use garage_rpc::*;
@@ -26,8 +30,9 @@ use garage_table::replication::{TableReplication, TableShardedReplication};
use garage_table::{PartitionKey, Table};
use crate::k2v::causality::*;
use crate::k2v::history_table::*;
use crate::k2v::item_table::*;
use crate::k2v::poll::*;
use crate::k2v::sub::*;
/// RPC messages for K2V
#[derive(Debug, Serialize, Deserialize)]
@@ -59,6 +64,8 @@ impl Rpc for K2VRpc {
pub struct K2VRpcHandler {
system: Arc<System>,
item_table: Arc<Table<K2VItemTable, TableShardedReplication>>,
history_table: Arc<Table<K2VHistoryTable, TableShardedReplication>>,
local_counter_tree: db::Tree,
endpoint: Arc<Endpoint<K2VRpc, Self>>,
subscriptions: Arc<SubscriptionManager>,
}
@@ -66,14 +73,21 @@ pub struct K2VRpcHandler {
impl K2VRpcHandler {
pub fn new(
system: Arc<System>,
db: &db::Db,
item_table: Arc<Table<K2VItemTable, TableShardedReplication>>,
history_table: Arc<Table<K2VHistoryTable, TableShardedReplication>>,
subscriptions: Arc<SubscriptionManager>,
) -> Arc<Self> {
let local_counter_tree = db
.open_tree("k2v_local_counter")
.expect("Unable to open DB tree for k2v local counter");
let endpoint = system.netapp.endpoint("garage_model/k2v/Rpc".to_string());
let rpc_handler = Arc::new(Self {
system,
item_table,
history_table,
local_counter_tree,
endpoint,
subscriptions,
});
@@ -181,7 +195,7 @@ impl K2VRpcHandler {
Ok(())
}
pub async fn poll(
pub async fn poll_item(
&self,
bucket_id: Uuid,
partition_key: String,
@@ -273,14 +287,18 @@ impl K2VRpcHandler {
}
fn local_insert(&self, item: &InsertedItem) -> Result<Option<K2VItem>, Error> {
let tree_key = self
.item_table
.data
.tree_key(&item.partition, &item.sort_key);
let now = now_msec();
self.item_table
.data
.update_entry_with(&tree_key[..], |ent| {
.update_entry_with(&item.partition, &item.sort_key, |tx, ent| {
let local_counter_key = item.partition.hash();
let old_local_counter = tx
.get(&self.local_counter_tree, &local_counter_key)?
.and_then(|x| x.try_into().ok())
.map(u64::from_be_bytes)
.unwrap_or_default();
let mut ent = ent.unwrap_or_else(|| {
K2VItem::new(
item.partition.bucket_id,
@@ -288,13 +306,37 @@ impl K2VRpcHandler {
item.sort_key.clone(),
)
});
ent.update(self.system.id, &item.causal_context, item.value.clone());
ent
let new_local_counter = ent.update(
self.system.id,
&item.causal_context,
item.value.clone(),
old_local_counter,
);
tx.insert(
&self.local_counter_tree,
&local_counter_key,
u64::to_be_bytes(new_local_counter),
)?;
let hist_entry = K2VHistoryEntry {
ins_item: ent.clone(),
node_counter: K2VHistorySortKey {
node: make_node_id(self.system.id),
counter: new_local_counter,
},
prev_counter: old_local_counter,
timestamp: now,
deleted: false.into(),
};
self.history_table.queue_insert(tx, &hist_entry)?;
Ok(ent)
})
}
async fn handle_poll(&self, key: &PollKey, ct: &CausalContext) -> Result<K2VItem, Error> {
let mut chan = self.subscriptions.subscribe(key);
async fn handle_poll_item(&self, key: &PollKey, ct: &CausalContext) -> Result<K2VItem, Error> {
let mut chan = self.subscriptions.subscribe_item(key);
let mut value = self
.item_table
@@ -331,7 +373,7 @@ impl EndpointHandler<K2VRpc> for K2VRpcHandler {
} => {
let delay = tokio::time::sleep(Duration::from_millis(*timeout_msec));
select! {
ret = self.handle_poll(key, causal_context) => ret.map(Some).map(K2VRpc::PollItemResponse),
ret = self.handle_poll_item(key, causal_context) => ret.map(Some).map(K2VRpc::PollItemResponse),
_ = delay => Ok(K2VRpc::PollItemResponse(None)),
}
}
+115
View File
@@ -0,0 +1,115 @@
use std::collections::HashMap;
use std::sync::Mutex;
use serde::{Deserialize, Serialize};
use tokio::sync::broadcast;
use crate::k2v::history_table::*;
use crate::k2v::item_table::*;
#[derive(Debug, Hash, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct PollKey {
pub partition: K2VItemPartition,
pub sort_key: String,
}
#[derive(Debug, Hash, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct PollRange {
pub partition: K2VItemPartition,
pub prefix: Option<String>,
pub start: Option<String>,
pub end: Option<String>,
}
#[derive(Default)]
pub struct SubscriptionManager {
item_subscriptions: Mutex<HashMap<PollKey, broadcast::Sender<K2VItem>>>,
range_subscriptions: Mutex<HashMap<PollRange, broadcast::Sender<K2VHistoryEntry>>>,
}
impl SubscriptionManager {
pub fn new() -> Self {
Self::default()
}
// ---- simple item polling ----
pub fn subscribe_item(&self, key: &PollKey) -> broadcast::Receiver<K2VItem> {
let mut subs = self.item_subscriptions.lock().unwrap();
if let Some(s) = subs.get(key) {
s.subscribe()
} else {
let (tx, rx) = broadcast::channel(8);
subs.insert(key.clone(), tx);
rx
}
}
pub fn notify_item(&self, item: &K2VItem) {
let key = PollKey {
partition: item.partition.clone(),
sort_key: item.sort_key.clone(),
};
let mut subs = self.item_subscriptions.lock().unwrap();
if let Some(s) = subs.get(&key) {
if s.send(item.clone()).is_err() {
// no more subscribers, remove channel from here
// (we will re-create it later if we need to subscribe again)
subs.remove(&key);
}
}
}
// ---- range polling ----
pub fn subscribe_range(&self, key: &PollRange) -> broadcast::Receiver<K2VHistoryEntry> {
let mut subs = self.range_subscriptions.lock().unwrap();
if let Some(s) = subs.get(key) {
s.subscribe()
} else {
let (tx, rx) = broadcast::channel(8);
subs.insert(key.clone(), tx);
rx
}
}
pub fn notify_range(&self, entry: &K2VHistoryEntry) {
let mut subs = self.range_subscriptions.lock().unwrap();
let mut dead_subs = vec![];
for (sub, chan) in subs.iter() {
if sub.matches(&entry) {
if chan.send(entry.clone()).is_err() {
dead_subs.push(sub.clone());
}
} else if chan.receiver_count() == 0 {
dead_subs.push(sub.clone());
}
}
for sub in dead_subs.iter() {
subs.remove(sub);
}
}
}
impl PollRange {
fn matches(&self, entry: &K2VHistoryEntry) -> bool {
entry.ins_item.partition == self.partition
&& self
.prefix
.as_ref()
.map(|x| entry.ins_item.sort_key.starts_with(x))
.unwrap_or(true)
&& self
.start
.as_ref()
.map(|x| entry.ins_item.sort_key >= *x)
.unwrap_or(true)
&& self
.end
.as_ref()
.map(|x| entry.ins_item.sort_key < *x)
.unwrap_or(true)
}
}
+102 -50
View File
@@ -1,45 +1,121 @@
use serde::{Deserialize, Serialize};
use garage_table::crdt::*;
use garage_table::*;
use garage_util::crdt::{self, Crdt};
use garage_util::data::*;
use garage_table::{DeletedFilter, EmptyKey, Entry, TableSchema};
use crate::permission::BucketKeyPerm;
use crate::prev::v051::key_table as old;
pub(crate) mod v05 {
use garage_util::crdt;
use serde::{Deserialize, Serialize};
/// An api key
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Key {
/// The id of the key (immutable), used as partition key
pub key_id: String,
/// An api key
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Key {
/// The id of the key (immutable), used as partition key
pub key_id: String,
/// Internal state of the key
pub state: crdt::Deletable<KeyParams>,
/// The secret_key associated
pub secret_key: String,
/// Name for the key
pub name: crdt::Lww<String>,
/// Is the key deleted
pub deleted: crdt::Bool,
/// Buckets in which the key is authorized. Empty if `Key` is deleted
// CRDT interaction: deleted implies authorized_buckets is empty
pub authorized_buckets: crdt::LwwMap<String, PermissionSet>,
}
/// Permission given to a key in a bucket
#[derive(PartialOrd, Ord, PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct PermissionSet {
/// The key can be used to read the bucket
pub allow_read: bool,
/// The key can be used to write in the bucket
pub allow_write: bool,
}
impl crdt::AutoCrdt for PermissionSet {
const WARN_IF_DIFFERENT: bool = true;
}
impl garage_util::migrate::InitialFormat for Key {}
}
/// Configuration for a key
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct KeyParams {
/// The secret_key associated (immutable)
pub secret_key: String,
mod v08 {
use super::v05;
use crate::permission::BucketKeyPerm;
use garage_util::crdt;
use garage_util::data::Uuid;
use serde::{Deserialize, Serialize};
/// Name for the key
pub name: crdt::Lww<String>,
/// An api key
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Key {
/// The id of the key (immutable), used as partition key
pub key_id: String,
/// Flag to allow users having this key to create buckets
pub allow_create_bucket: crdt::Lww<bool>,
/// Internal state of the key
pub state: crdt::Deletable<KeyParams>,
}
/// If the key is present: it gives some permissions,
/// a map of bucket IDs (uuids) to permissions.
/// Otherwise no permissions are granted to key
pub authorized_buckets: crdt::Map<Uuid, BucketKeyPerm>,
/// Configuration for a key
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct KeyParams {
/// The secret_key associated (immutable)
pub secret_key: String,
/// A key can have a local view of buckets names it is
/// the only one to see, this is the namespace for these aliases
pub local_aliases: crdt::LwwMap<String, Option<Uuid>>,
/// Name for the key
pub name: crdt::Lww<String>,
/// Flag to allow users having this key to create buckets
pub allow_create_bucket: crdt::Lww<bool>,
/// If the key is present: it gives some permissions,
/// a map of bucket IDs (uuids) to permissions.
/// Otherwise no permissions are granted to key
pub authorized_buckets: crdt::Map<Uuid, BucketKeyPerm>,
/// A key can have a local view of buckets names it is
/// the only one to see, this is the namespace for these aliases
pub local_aliases: crdt::LwwMap<String, Option<Uuid>>,
}
impl garage_util::migrate::Migrate for Key {
type Previous = v05::Key;
fn migrate(old_k: v05::Key) -> Key {
let name = crdt::Lww::raw(old_k.name.timestamp(), old_k.name.get().clone());
let state = if old_k.deleted.get() {
crdt::Deletable::Deleted
} else {
// Authorized buckets is ignored here,
// migration is performed in specific migration code in
// garage/migrate.rs
crdt::Deletable::Present(KeyParams {
secret_key: old_k.secret_key,
name,
allow_create_bucket: crdt::Lww::new(false),
authorized_buckets: crdt::Map::new(),
local_aliases: crdt::LwwMap::new(),
})
};
Key {
key_id: old_k.key_id,
state,
}
}
}
}
pub use v08::*;
impl KeyParams {
fn new(secret_key: &str, name: &str) -> Self {
KeyParams {
@@ -173,28 +249,4 @@ impl TableSchema for KeyTable {
}
}
}
fn try_migrate(bytes: &[u8]) -> Option<Self::E> {
let old_k = rmp_serde::decode::from_read_ref::<_, old::Key>(bytes).ok()?;
let name = crdt::Lww::raw(old_k.name.timestamp(), old_k.name.get().clone());
let state = if old_k.deleted.get() {
crdt::Deletable::Deleted
} else {
// Authorized buckets is ignored here,
// migration is performed in specific migration code in
// garage/migrate.rs
crdt::Deletable::Present(KeyParams {
secret_key: old_k.secret_key,
name,
allow_create_bucket: crdt::Lww::new(false),
authorized_buckets: crdt::Map::new(),
local_aliases: crdt::LwwMap::new(),
})
};
Some(Key {
key_id: old_k.key_id,
state,
})
}
}
+3 -2
View File
@@ -2,6 +2,7 @@ use std::sync::Arc;
use garage_util::crdt::*;
use garage_util::data::*;
use garage_util::encode::nonversioned_decode;
use garage_util::error::Error as GarageError;
use garage_util::time::*;
@@ -28,8 +29,8 @@ impl Migrate {
let mut old_buckets = vec![];
for res in tree.iter().map_err(GarageError::from)? {
let (_k, v) = res.map_err(GarageError::from)?;
let bucket = rmp_serde::decode::from_read_ref::<_, old_bucket::Bucket>(&v[..])
.map_err(GarageError::from)?;
let bucket =
nonversioned_decode::<old_bucket::Bucket>(&v[..]).map_err(GarageError::from)?;
old_buckets.push(bucket);
}
+1 -1
View File
@@ -3,7 +3,7 @@ use serde::{Deserialize, Serialize};
use garage_table::crdt::Crdt;
use garage_table::*;
use super::key_table::PermissionSet;
use crate::key_table::v05::PermissionSet;
/// A bucket is a collection of objects
///
-50
View File
@@ -1,50 +0,0 @@
use serde::{Deserialize, Serialize};
use garage_table::crdt::*;
use garage_table::*;
/// An api key
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Key {
/// The id of the key (immutable), used as partition key
pub key_id: String,
/// The secret_key associated
pub secret_key: String,
/// Name for the key
pub name: crdt::Lww<String>,
/// Is the key deleted
pub deleted: crdt::Bool,
/// Buckets in which the key is authorized. Empty if `Key` is deleted
// CRDT interaction: deleted implies authorized_buckets is empty
pub authorized_buckets: crdt::LwwMap<String, PermissionSet>,
}
/// Permission given to a key in a bucket
#[derive(PartialOrd, Ord, PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct PermissionSet {
/// The key can be used to read the bucket
pub allow_read: bool,
/// The key can be used to write in the bucket
pub allow_write: bool,
}
impl AutoCrdt for PermissionSet {
const WARN_IF_DIFFERENT: bool = true;
}
impl Crdt for Key {
fn merge(&mut self, other: &Self) {
self.name.merge(&other.name);
self.deleted.merge(&other.deleted);
if self.deleted.get() {
self.authorized_buckets.clear();
} else {
self.authorized_buckets.merge(&other.authorized_buckets);
}
}
}
-3
View File
@@ -1,4 +1 @@
pub(crate) mod bucket_table;
pub(crate) mod key_table;
pub(crate) mod object_table;
pub(crate) mod version_table;
-149
View File
@@ -1,149 +0,0 @@
use serde::{Deserialize, Serialize};
use std::collections::BTreeMap;
use garage_util::data::*;
use garage_table::crdt::*;
/// An object
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Object {
/// The bucket in which the object is stored, used as partition key
pub bucket: String,
/// The key at which the object is stored in its bucket, used as sorting key
pub key: String,
/// The list of currenty stored versions of the object
versions: Vec<ObjectVersion>,
}
impl Object {
/// Get a list of currently stored versions of `Object`
pub fn versions(&self) -> &[ObjectVersion] {
&self.versions[..]
}
}
/// Informations about a version of an object
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct ObjectVersion {
/// Id of the version
pub uuid: Uuid,
/// Timestamp of when the object was created
pub timestamp: u64,
/// State of the version
pub state: ObjectVersionState,
}
/// State of an object version
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub enum ObjectVersionState {
/// The version is being received
Uploading(ObjectVersionHeaders),
/// The version is fully received
Complete(ObjectVersionData),
/// The version uploaded containded errors or the upload was explicitly aborted
Aborted,
}
impl Crdt for ObjectVersionState {
fn merge(&mut self, other: &Self) {
use ObjectVersionState::*;
match other {
Aborted => {
*self = Aborted;
}
Complete(b) => match self {
Aborted => {}
Complete(a) => {
a.merge(b);
}
Uploading(_) => {
*self = Complete(b.clone());
}
},
Uploading(_) => {}
}
}
}
/// Data stored in object version
#[derive(PartialEq, Eq, PartialOrd, Ord, Clone, Debug, Serialize, Deserialize)]
pub enum ObjectVersionData {
/// The object was deleted, this Version is a tombstone to mark it as such
DeleteMarker,
/// The object is short, it's stored inlined
Inline(ObjectVersionMeta, #[serde(with = "serde_bytes")] Vec<u8>),
/// The object is not short, Hash of first block is stored here, next segments hashes are
/// stored in the version table
FirstBlock(ObjectVersionMeta, Hash),
}
impl AutoCrdt for ObjectVersionData {
const WARN_IF_DIFFERENT: bool = true;
}
/// Metadata about the object version
#[derive(PartialEq, Eq, PartialOrd, Ord, Clone, Debug, Serialize, Deserialize)]
pub struct ObjectVersionMeta {
/// Headers to send to the client
pub headers: ObjectVersionHeaders,
/// Size of the object
pub size: u64,
/// etag of the object
pub etag: String,
}
/// Additional headers for an object
#[derive(PartialEq, Eq, PartialOrd, Ord, Clone, Debug, Serialize, Deserialize)]
pub struct ObjectVersionHeaders {
/// Content type of the object
pub content_type: String,
/// Any other http headers to send
pub other: BTreeMap<String, String>,
}
impl ObjectVersion {
fn cmp_key(&self) -> (u64, Uuid) {
(self.timestamp, self.uuid)
}
/// Is the object version completely received
pub fn is_complete(&self) -> bool {
matches!(self.state, ObjectVersionState::Complete(_))
}
}
impl Crdt for Object {
fn merge(&mut self, other: &Self) {
// Merge versions from other into here
for other_v in other.versions.iter() {
match self
.versions
.binary_search_by(|v| v.cmp_key().cmp(&other_v.cmp_key()))
{
Ok(i) => {
self.versions[i].state.merge(&other_v.state);
}
Err(i) => {
self.versions.insert(i, other_v.clone());
}
}
}
// Remove versions which are obsolete, i.e. those that come
// before the last version which .is_complete().
let last_complete = self
.versions
.iter()
.enumerate()
.rev()
.find(|(_, v)| v.is_complete())
.map(|(vi, _)| vi);
if let Some(last_vi) = last_complete {
self.versions = self.versions.drain(last_vi..).collect::<Vec<_>>();
}
}
}
-79
View File
@@ -1,79 +0,0 @@
use serde::{Deserialize, Serialize};
use garage_util::data::*;
use garage_table::crdt::*;
use garage_table::*;
/// A version of an object
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Version {
/// UUID of the version, used as partition key
pub uuid: Uuid,
// Actual data: the blocks for this version
// In the case of a multipart upload, also store the etags
// of individual parts and check them when doing CompleteMultipartUpload
/// Is this version deleted
pub deleted: crdt::Bool,
/// list of blocks of data composing the version
pub blocks: crdt::Map<VersionBlockKey, VersionBlock>,
/// Etag of each part in case of a multipart upload, empty otherwise
pub parts_etags: crdt::Map<u64, String>,
// Back link to bucket+key so that we can figure if
// this was deleted later on
/// Bucket in which the related object is stored
pub bucket: String,
/// Key in which the related object is stored
pub key: String,
}
#[derive(PartialEq, Eq, Clone, Copy, Debug, Serialize, Deserialize)]
pub struct VersionBlockKey {
/// Number of the part
pub part_number: u64,
/// Offset of this sub-segment in its part
pub offset: u64,
}
impl Ord for VersionBlockKey {
fn cmp(&self, other: &Self) -> std::cmp::Ordering {
self.part_number
.cmp(&other.part_number)
.then(self.offset.cmp(&other.offset))
}
}
impl PartialOrd for VersionBlockKey {
fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
Some(self.cmp(other))
}
}
/// Informations about a single block
#[derive(PartialEq, Eq, Ord, PartialOrd, Clone, Copy, Debug, Serialize, Deserialize)]
pub struct VersionBlock {
/// Blake2 sum of the block
pub hash: Hash,
/// Size of the block
pub size: u64,
}
impl AutoCrdt for VersionBlock {
const WARN_IF_DIFFERENT: bool = true;
}
impl Crdt for Version {
fn merge(&mut self, other: &Self) {
self.deleted.merge(&other.deleted);
if self.deleted.get() {
self.blocks.clear();
self.parts_etags.clear();
} else {
self.blocks.merge(&other.blocks);
self.parts_etags.merge(&other.parts_etags);
}
}
}
+19 -10
View File
@@ -1,4 +1,3 @@
use serde::{Deserialize, Serialize};
use std::sync::Arc;
use garage_db as db;
@@ -10,19 +9,29 @@ use garage_table::*;
use garage_block::manager::*;
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct BlockRef {
/// Hash (blake2 sum) of the block, used as partition key
pub block: Hash,
mod v08 {
use garage_util::crdt;
use garage_util::data::{Hash, Uuid};
use serde::{Deserialize, Serialize};
/// Id of the Version for the object containing this block, used as sorting key
pub version: Uuid,
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct BlockRef {
/// Hash (blake2 sum) of the block, used as partition key
pub block: Hash,
// Keep track of deleted status
/// Is the Version that contains this block deleted
pub deleted: crdt::Bool,
/// Id of the Version for the object containing this block, used as sorting key
pub version: Uuid,
// Keep track of deleted status
/// Is the Version that contains this block deleted
pub deleted: crdt::Bool,
}
impl garage_util::migrate::InitialFormat for BlockRef {}
}
pub use v08::*;
impl Entry<Hash, Uuid> for BlockRef {
fn partition_key(&self) -> &Hash {
&self.block
+137 -159
View File
@@ -1,10 +1,8 @@
use serde::{Deserialize, Serialize};
use std::collections::BTreeMap;
use std::sync::Arc;
use garage_db as db;
use garage_util::background::BackgroundRunner;
use garage_util::data::*;
use garage_table::crdt::*;
@@ -14,25 +12,126 @@ use garage_table::*;
use crate::index_counter::*;
use crate::s3::version_table::*;
use crate::prev::v051::object_table as old;
pub const OBJECTS: &str = "objects";
pub const UNFINISHED_UPLOADS: &str = "unfinished_uploads";
pub const BYTES: &str = "bytes";
/// An object
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Object {
/// The bucket in which the object is stored, used as partition key
pub bucket_id: Uuid,
mod v05 {
use garage_util::data::{Hash, Uuid};
use serde::{Deserialize, Serialize};
use std::collections::BTreeMap;
/// The key at which the object is stored in its bucket, used as sorting key
pub key: String,
/// An object
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Object {
/// The bucket in which the object is stored, used as partition key
pub bucket: String,
/// The list of currenty stored versions of the object
versions: Vec<ObjectVersion>,
/// The key at which the object is stored in its bucket, used as sorting key
pub key: String,
/// The list of currenty stored versions of the object
pub(super) versions: Vec<ObjectVersion>,
}
/// Informations about a version of an object
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct ObjectVersion {
/// Id of the version
pub uuid: Uuid,
/// Timestamp of when the object was created
pub timestamp: u64,
/// State of the version
pub state: ObjectVersionState,
}
/// State of an object version
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub enum ObjectVersionState {
/// The version is being received
Uploading(ObjectVersionHeaders),
/// The version is fully received
Complete(ObjectVersionData),
/// The version uploaded containded errors or the upload was explicitly aborted
Aborted,
}
/// Data stored in object version
#[derive(PartialEq, Eq, PartialOrd, Ord, Clone, Debug, Serialize, Deserialize)]
pub enum ObjectVersionData {
/// The object was deleted, this Version is a tombstone to mark it as such
DeleteMarker,
/// The object is short, it's stored inlined
Inline(ObjectVersionMeta, #[serde(with = "serde_bytes")] Vec<u8>),
/// The object is not short, Hash of first block is stored here, next segments hashes are
/// stored in the version table
FirstBlock(ObjectVersionMeta, Hash),
}
/// Metadata about the object version
#[derive(PartialEq, Eq, PartialOrd, Ord, Clone, Debug, Serialize, Deserialize)]
pub struct ObjectVersionMeta {
/// Headers to send to the client
pub headers: ObjectVersionHeaders,
/// Size of the object
pub size: u64,
/// etag of the object
pub etag: String,
}
/// Additional headers for an object
#[derive(PartialEq, Eq, PartialOrd, Ord, Clone, Debug, Serialize, Deserialize)]
pub struct ObjectVersionHeaders {
/// Content type of the object
pub content_type: String,
/// Any other http headers to send
pub other: BTreeMap<String, String>,
}
impl garage_util::migrate::InitialFormat for Object {}
}
mod v08 {
use garage_util::data::Uuid;
use serde::{Deserialize, Serialize};
use super::v05;
pub use v05::{
ObjectVersion, ObjectVersionData, ObjectVersionHeaders, ObjectVersionMeta,
ObjectVersionState,
};
/// An object
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Object {
/// The bucket in which the object is stored, used as partition key
pub bucket_id: Uuid,
/// The key at which the object is stored in its bucket, used as sorting key
pub key: String,
/// The list of currenty stored versions of the object
pub(super) versions: Vec<ObjectVersion>,
}
impl garage_util::migrate::Migrate for Object {
type Previous = v05::Object;
fn migrate(old: v05::Object) -> Object {
use garage_util::data::blake2sum;
Object {
bucket_id: blake2sum(old.bucket.as_bytes()),
key: old.key,
versions: old.versions,
}
}
}
}
pub use v08::*;
impl Object {
/// Initialize an Object struct from parts
pub fn new(bucket_id: Uuid, key: String, versions: Vec<ObjectVersion>) -> Self {
@@ -69,28 +168,6 @@ impl Object {
}
}
/// Informations about a version of an object
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct ObjectVersion {
/// Id of the version
pub uuid: Uuid,
/// Timestamp of when the object was created
pub timestamp: u64,
/// State of the version
pub state: ObjectVersionState,
}
/// State of an object version
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub enum ObjectVersionState {
/// The version is being received
Uploading(ObjectVersionHeaders),
/// The version is fully received
Complete(ObjectVersionData),
/// The version uploaded containded errors or the upload was explicitly aborted
Aborted,
}
impl Crdt for ObjectVersionState {
fn merge(&mut self, other: &Self) {
use ObjectVersionState::*;
@@ -112,42 +189,10 @@ impl Crdt for ObjectVersionState {
}
}
/// Data stored in object version
#[derive(PartialEq, Eq, PartialOrd, Ord, Clone, Debug, Serialize, Deserialize)]
pub enum ObjectVersionData {
/// The object was deleted, this Version is a tombstone to mark it as such
DeleteMarker,
/// The object is short, it's stored inlined
Inline(ObjectVersionMeta, #[serde(with = "serde_bytes")] Vec<u8>),
/// The object is not short, Hash of first block is stored here, next segments hashes are
/// stored in the version table
FirstBlock(ObjectVersionMeta, Hash),
}
impl AutoCrdt for ObjectVersionData {
const WARN_IF_DIFFERENT: bool = true;
}
/// Metadata about the object version
#[derive(PartialEq, Eq, PartialOrd, Ord, Clone, Debug, Serialize, Deserialize)]
pub struct ObjectVersionMeta {
/// Headers to send to the client
pub headers: ObjectVersionHeaders,
/// Size of the object
pub size: u64,
/// etag of the object
pub etag: String,
}
/// Additional headers for an object
#[derive(PartialEq, Eq, PartialOrd, Ord, Clone, Debug, Serialize, Deserialize)]
pub struct ObjectVersionHeaders {
/// Content type of the object
pub content_type: String,
/// Any other http headers to send
pub other: BTreeMap<String, String>,
}
impl ObjectVersion {
fn cmp_key(&self) -> (u64, Uuid) {
(self.timestamp, self.uuid)
@@ -221,7 +266,6 @@ impl Crdt for Object {
}
pub struct ObjectTable {
pub background: Arc<BackgroundRunner>,
pub version_table: Arc<Table<VersionTable, TableShardedReplication>>,
pub object_counter_table: Arc<IndexCounter<Object>>,
}
@@ -255,34 +299,34 @@ impl TableSchema for ObjectTable {
);
}
// 2. Spawn threads that propagates deletions to version table
let version_table = self.version_table.clone();
let old = old.cloned();
let new = new.cloned();
self.background.spawn(async move {
if let (Some(old_v), Some(new_v)) = (old, new) {
// Propagate deletion of old versions
for v in old_v.versions.iter() {
let newly_deleted = match new_v
.versions
.binary_search_by(|nv| nv.cmp_key().cmp(&v.cmp_key()))
{
Err(_) => true,
Ok(i) => {
new_v.versions[i].state == ObjectVersionState::Aborted
&& v.state != ObjectVersionState::Aborted
}
};
if newly_deleted {
let deleted_version =
Version::new(v.uuid, old_v.bucket_id, old_v.key.clone(), true);
version_table.insert(&deleted_version).await?;
// 2. Enqueue propagation deletions to version table
if let (Some(old_v), Some(new_v)) = (old, new) {
// Propagate deletion of old versions
for v in old_v.versions.iter() {
let newly_deleted = match new_v
.versions
.binary_search_by(|nv| nv.cmp_key().cmp(&v.cmp_key()))
{
Err(_) => true,
Ok(i) => {
new_v.versions[i].state == ObjectVersionState::Aborted
&& v.state != ObjectVersionState::Aborted
}
};
if newly_deleted {
let deleted_version =
Version::new(v.uuid, old_v.bucket_id, old_v.key.clone(), true);
let res = self.version_table.queue_insert(tx, &deleted_version);
if let Err(e) = db::unabort(res)? {
error!(
"Unable to enqueue version deletion propagation: {}. A repair will be needed.",
e
);
}
}
}
Ok(())
});
}
Ok(())
}
@@ -292,11 +336,6 @@ impl TableSchema for ObjectTable {
ObjectFilter::IsUploading => entry.versions.iter().any(|v| v.is_uploading()),
}
}
fn try_migrate(bytes: &[u8]) -> Option<Self::E> {
let old_obj = rmp_serde::decode::from_read_ref::<_, old::Object>(bytes).ok()?;
Some(migrate_object(old_obj))
}
}
impl CountedItem for Object {
@@ -341,64 +380,3 @@ impl CountedItem for Object {
]
}
}
// vvvvvvvv migration code, stupid stuff vvvvvvvvvvvv
// (we just want to change bucket into bucket_id by hashing it)
fn migrate_object(o: old::Object) -> Object {
let versions = o
.versions()
.iter()
.cloned()
.map(migrate_object_version)
.collect();
Object {
bucket_id: blake2sum(o.bucket.as_bytes()),
key: o.key,
versions,
}
}
fn migrate_object_version(v: old::ObjectVersion) -> ObjectVersion {
ObjectVersion {
uuid: Uuid::try_from(v.uuid.as_slice()).unwrap(),
timestamp: v.timestamp,
state: match v.state {
old::ObjectVersionState::Uploading(h) => {
ObjectVersionState::Uploading(migrate_object_version_headers(h))
}
old::ObjectVersionState::Complete(d) => {
ObjectVersionState::Complete(migrate_object_version_data(d))
}
old::ObjectVersionState::Aborted => ObjectVersionState::Aborted,
},
}
}
fn migrate_object_version_headers(h: old::ObjectVersionHeaders) -> ObjectVersionHeaders {
ObjectVersionHeaders {
content_type: h.content_type,
other: h.other,
}
}
fn migrate_object_version_data(d: old::ObjectVersionData) -> ObjectVersionData {
match d {
old::ObjectVersionData::DeleteMarker => ObjectVersionData::DeleteMarker,
old::ObjectVersionData::Inline(m, b) => {
ObjectVersionData::Inline(migrate_object_version_meta(m), b)
}
old::ObjectVersionData::FirstBlock(m, h) => ObjectVersionData::FirstBlock(
migrate_object_version_meta(m),
Hash::try_from(h.as_slice()).unwrap(),
),
}
}
fn migrate_object_version_meta(m: old::ObjectVersionMeta) -> ObjectVersionMeta {
ObjectVersionMeta {
headers: migrate_object_version_headers(m.headers),
size: m.size,
etag: m.etag,
}
}
+112 -101
View File
@@ -1,9 +1,7 @@
use serde::{Deserialize, Serialize};
use std::sync::Arc;
use garage_db as db;
use garage_util::background::BackgroundRunner;
use garage_util::data::*;
use garage_table::crdt::*;
@@ -12,32 +10,108 @@ use garage_table::*;
use crate::s3::block_ref_table::*;
use crate::prev::v051::version_table as old;
mod v05 {
use garage_util::crdt;
use garage_util::data::{Hash, Uuid};
use serde::{Deserialize, Serialize};
/// A version of an object
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Version {
/// UUID of the version, used as partition key
pub uuid: Uuid,
/// A version of an object
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Version {
/// UUID of the version, used as partition key
pub uuid: Uuid,
// Actual data: the blocks for this version
// In the case of a multipart upload, also store the etags
// of individual parts and check them when doing CompleteMultipartUpload
/// Is this version deleted
pub deleted: crdt::Bool,
/// list of blocks of data composing the version
pub blocks: crdt::Map<VersionBlockKey, VersionBlock>,
/// Etag of each part in case of a multipart upload, empty otherwise
pub parts_etags: crdt::Map<u64, String>,
// Actual data: the blocks for this version
// In the case of a multipart upload, also store the etags
// of individual parts and check them when doing CompleteMultipartUpload
/// Is this version deleted
pub deleted: crdt::Bool,
/// list of blocks of data composing the version
pub blocks: crdt::Map<VersionBlockKey, VersionBlock>,
/// Etag of each part in case of a multipart upload, empty otherwise
pub parts_etags: crdt::Map<u64, String>,
// Back link to bucket+key so that we can figure if
// this was deleted later on
/// Bucket in which the related object is stored
pub bucket_id: Uuid,
/// Key in which the related object is stored
pub key: String,
// Back link to bucket+key so that we can figure if
// this was deleted later on
/// Bucket in which the related object is stored
pub bucket: String,
/// Key in which the related object is stored
pub key: String,
}
#[derive(PartialEq, Eq, Clone, Copy, Debug, Serialize, Deserialize)]
pub struct VersionBlockKey {
/// Number of the part
pub part_number: u64,
/// Offset of this sub-segment in its part
pub offset: u64,
}
/// Informations about a single block
#[derive(PartialEq, Eq, Ord, PartialOrd, Clone, Copy, Debug, Serialize, Deserialize)]
pub struct VersionBlock {
/// Blake2 sum of the block
pub hash: Hash,
/// Size of the block
pub size: u64,
}
impl garage_util::migrate::InitialFormat for Version {}
}
mod v08 {
use garage_util::crdt;
use garage_util::data::Uuid;
use serde::{Deserialize, Serialize};
use super::v05;
/// A version of an object
#[derive(PartialEq, Eq, Clone, Debug, Serialize, Deserialize)]
pub struct Version {
/// UUID of the version, used as partition key
pub uuid: Uuid,
// Actual data: the blocks for this version
// In the case of a multipart upload, also store the etags
// of individual parts and check them when doing CompleteMultipartUpload
/// Is this version deleted
pub deleted: crdt::Bool,
/// list of blocks of data composing the version
pub blocks: crdt::Map<VersionBlockKey, VersionBlock>,
/// Etag of each part in case of a multipart upload, empty otherwise
pub parts_etags: crdt::Map<u64, String>,
// Back link to bucket+key so that we can figure if
// this was deleted later on
/// Bucket in which the related object is stored
pub bucket_id: Uuid,
/// Key in which the related object is stored
pub key: String,
}
pub use v05::{VersionBlock, VersionBlockKey};
impl garage_util::migrate::Migrate for Version {
type Previous = v05::Version;
fn migrate(old: v05::Version) -> Version {
use garage_util::data::blake2sum;
Version {
uuid: old.uuid,
deleted: old.deleted,
blocks: old.blocks,
parts_etags: old.parts_etags,
bucket_id: blake2sum(old.bucket.as_bytes()),
key: old.key,
}
}
}
}
pub use v08::*;
impl Version {
pub fn new(uuid: Uuid, bucket_id: Uuid, key: String, deleted: bool) -> Self {
Self {
@@ -65,14 +139,6 @@ impl Version {
}
}
#[derive(PartialEq, Eq, Clone, Copy, Debug, Serialize, Deserialize)]
pub struct VersionBlockKey {
/// Number of the part
pub part_number: u64,
/// Offset of this sub-segment in its part
pub offset: u64,
}
impl Ord for VersionBlockKey {
fn cmp(&self, other: &Self) -> std::cmp::Ordering {
self.part_number
@@ -87,15 +153,6 @@ impl PartialOrd for VersionBlockKey {
}
}
/// Informations about a single block
#[derive(PartialEq, Eq, Ord, PartialOrd, Clone, Copy, Debug, Serialize, Deserialize)]
pub struct VersionBlock {
/// Blake2 sum of the block
pub hash: Hash,
/// Size of the block
pub size: u64,
}
impl AutoCrdt for VersionBlock {
const WARN_IF_DIFFERENT: bool = true;
}
@@ -127,7 +184,6 @@ impl Crdt for Version {
}
pub struct VersionTable {
pub background: Arc<BackgroundRunner>,
pub block_ref_table: Arc<Table<BlockRefTable, TableShardedReplication>>,
}
@@ -141,33 +197,26 @@ impl TableSchema for VersionTable {
fn updated(
&self,
_tx: &mut db::Transaction,
tx: &mut db::Transaction,
old: Option<&Self::E>,
new: Option<&Self::E>,
) -> db::TxOpResult<()> {
let block_ref_table = self.block_ref_table.clone();
let old = old.cloned();
let new = new.cloned();
self.background.spawn(async move {
if let (Some(old_v), Some(new_v)) = (old, new) {
// Propagate deletion of version blocks
if new_v.deleted.get() && !old_v.deleted.get() {
let deleted_block_refs = old_v
.blocks
.items()
.iter()
.map(|(_k, vb)| BlockRef {
block: vb.hash,
version: old_v.uuid,
deleted: true.into(),
})
.collect::<Vec<_>>();
block_ref_table.insert_many(&deleted_block_refs[..]).await?;
if let (Some(old_v), Some(new_v)) = (old, new) {
// Propagate deletion of version blocks
if new_v.deleted.get() && !old_v.deleted.get() {
let deleted_block_refs = old_v.blocks.items().iter().map(|(_k, vb)| BlockRef {
block: vb.hash,
version: old_v.uuid,
deleted: true.into(),
});
for block_ref in deleted_block_refs {
let res = self.block_ref_table.queue_insert(tx, &block_ref);
if let Err(e) = db::unabort(res)? {
error!("Unable to enqueue block ref deletion propagation: {}. A repair will be needed.", e);
}
}
}
Ok(())
});
}
Ok(())
}
@@ -175,42 +224,4 @@ impl TableSchema for VersionTable {
fn matches_filter(entry: &Self::E, filter: &Self::Filter) -> bool {
filter.apply(entry.deleted.get())
}
fn try_migrate(bytes: &[u8]) -> Option<Self::E> {
let old = rmp_serde::decode::from_read_ref::<_, old::Version>(bytes).ok()?;
let blocks = old
.blocks
.items()
.iter()
.map(|(k, v)| {
(
VersionBlockKey {
part_number: k.part_number,
offset: k.offset,
},
VersionBlock {
hash: Hash::try_from(v.hash.as_slice()).unwrap(),
size: v.size,
},
)
})
.collect::<crdt::Map<_, _>>();
let parts_etags = old
.parts_etags
.items()
.iter()
.map(|(k, v)| (*k, v.clone()))
.collect::<crdt::Map<_, _>>();
Some(Version {
uuid: Hash::try_from(old.uuid.as_slice()).unwrap(),
deleted: crdt::Bool::new(old.deleted.get()),
blocks,
parts_etags,
bucket_id: blake2sum(old.bucket.as_bytes()),
key: old.key,
})
}
}
+2 -3
View File
@@ -1,6 +1,6 @@
[package]
name = "garage_rpc"
version = "0.8.0"
version = "0.8.1"
authors = ["Alex Auvolat <alex@adnab.me>"]
edition = "2018"
license = "AGPL-3.0"
@@ -14,7 +14,7 @@ path = "lib.rs"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
garage_util = { version = "0.8.0", path = "../util" }
garage_util = { version = "0.8.1", path = "../util" }
arc-swap = "1.0"
bytes = "1.0"
@@ -25,7 +25,6 @@ rand = "0.8"
sodiumoxide = { version = "0.2.5-0", package = "kuska-sodiumoxide" }
async-trait = "0.1.7"
rmp-serde = "0.15"
serde = { version = "1.0", default-features = false, features = ["derive", "rc"] }
serde_bytes = "0.11"
serde_json = "1.0"
+13 -8
View File
@@ -5,6 +5,7 @@ use serde::{Deserialize, Serialize};
use garage_util::crdt::{AutoCrdt, Crdt, LwwMap};
use garage_util::data::*;
use garage_util::encode::nonversioned_encode;
use garage_util::error::*;
use crate::ring::*;
@@ -35,6 +36,8 @@ pub struct ClusterLayout {
pub staging_hash: Hash,
}
impl garage_util::migrate::InitialFormat for ClusterLayout {}
#[derive(PartialEq, Eq, PartialOrd, Ord, Clone, Debug, Serialize, Deserialize)]
pub struct NodeRoleV(pub Option<NodeRole>);
@@ -68,7 +71,7 @@ impl NodeRole {
impl ClusterLayout {
pub fn new(replication_factor: usize) -> Self {
let empty_lwwmap = LwwMap::new();
let empty_lwwmap_hash = blake2sum(&rmp_to_vec_all_named(&empty_lwwmap).unwrap()[..]);
let empty_lwwmap_hash = blake2sum(&nonversioned_encode(&empty_lwwmap).unwrap()[..]);
ClusterLayout {
version: 0,
@@ -90,7 +93,7 @@ impl ClusterLayout {
Ordering::Equal => {
self.staging.merge(&other.staging);
let new_staging_hash = blake2sum(&rmp_to_vec_all_named(&self.staging).unwrap()[..]);
let new_staging_hash = blake2sum(&nonversioned_encode(&self.staging).unwrap()[..]);
let changed = new_staging_hash != self.staging_hash;
self.staging_hash = new_staging_hash;
@@ -125,7 +128,7 @@ To know the correct value of the new layout version, invoke `garage layout show`
}
self.staging.clear();
self.staging_hash = blake2sum(&rmp_to_vec_all_named(&self.staging).unwrap()[..]);
self.staging_hash = blake2sum(&nonversioned_encode(&self.staging).unwrap()[..]);
self.version += 1;
@@ -149,7 +152,7 @@ To know the correct value of the new layout version, invoke `garage layout show`
}
self.staging.clear();
self.staging_hash = blake2sum(&rmp_to_vec_all_named(&self.staging).unwrap()[..]);
self.staging_hash = blake2sum(&nonversioned_encode(&self.staging).unwrap()[..]);
self.version += 1;
@@ -178,7 +181,7 @@ To know the correct value of the new layout version, invoke `garage layout show`
/// returns true if consistent, false if error
pub fn check(&self) -> bool {
// Check that the hash of the staging data is correct
let staging_hash = blake2sum(&rmp_to_vec_all_named(&self.staging).unwrap()[..]);
let staging_hash = blake2sum(&nonversioned_encode(&self.staging).unwrap()[..]);
if staging_hash != self.staging_hash {
return false;
}
@@ -254,9 +257,11 @@ To know the correct value of the new layout version, invoke `garage layout show`
match self.initial_partition_assignation() {
Some(initial_partitions) => {
for (part, ipart) in partitions.iter_mut().zip(initial_partitions.iter()) {
for (id, info) in ipart.nodes.iter() {
if part.nodes.len() < self.replication_factor {
part.add(None, n_zones, id, info.unwrap());
for _ in 0..2 {
for (id, info) in ipart.nodes.iter() {
if part.nodes.len() < self.replication_factor {
part.add(None, n_zones, id, info.unwrap());
}
}
}
assert!(part.nodes.len() == self.replication_factor);
+1
View File
@@ -9,6 +9,7 @@ mod consul;
mod kubernetes;
pub mod layout;
pub mod replication_mode;
pub mod ring;
pub mod system;
@@ -1,3 +1,4 @@
#[derive(Clone, Copy)]
pub enum ReplicationMode {
None,
TwoWay,
+5 -13
View File
@@ -5,7 +5,6 @@ use std::time::Duration;
use futures::future::join_all;
use futures::stream::futures_unordered::FuturesUnordered;
use futures::stream::StreamExt;
use futures_util::future::FutureExt;
use tokio::select;
use tokio::sync::watch;
@@ -24,7 +23,6 @@ pub use netapp::message::{
use netapp::peering::fullmesh::FullMeshPeeringStrategy;
pub use netapp::{self, NetApp, NodeID};
use garage_util::background::BackgroundRunner;
use garage_util::data::*;
use garage_util::error::Error;
use garage_util::metrics::RecordDuration;
@@ -94,7 +92,6 @@ pub struct RpcHelper(Arc<RpcHelperInner>);
struct RpcHelperInner {
our_node_id: Uuid,
fullmesh: Arc<FullMeshPeeringStrategy>,
background: Arc<BackgroundRunner>,
ring: watch::Receiver<Arc<Ring>>,
metrics: RpcMetrics,
rpc_timeout: Duration,
@@ -104,7 +101,6 @@ impl RpcHelper {
pub(crate) fn new(
our_node_id: Uuid,
fullmesh: Arc<FullMeshPeeringStrategy>,
background: Arc<BackgroundRunner>,
ring: watch::Receiver<Arc<Ring>>,
rpc_timeout: Option<Duration>,
) -> Self {
@@ -113,7 +109,6 @@ impl RpcHelper {
Self(Arc::new(RpcHelperInner {
our_node_id,
fullmesh,
background,
ring,
metrics,
rpc_timeout: rpc_timeout.unwrap_or(DEFAULT_TIMEOUT),
@@ -377,16 +372,13 @@ impl RpcHelper {
if !resp_stream.is_empty() {
// Continue remaining requests in background.
// Continue the remaining requests immediately using tokio::spawn
// but enqueue a task in the background runner
// to ensure that the process won't exit until the requests are done
// (if we had just enqueued the resp_stream.collect directly in the background runner,
// the requests might have been put on hold in the background runner's queue,
// in which case they might timeout or otherwise fail)
let wait_finished_fut = tokio::spawn(async move {
// Note: these requests can get interrupted on process shutdown,
// we must not count on them being executed for certain.
// For all background things that have to happen with certainty,
// they have to be put in a proper queue that is persisted to disk.
tokio::spawn(async move {
resp_stream.collect::<Vec<Result<_, _>>>().await;
});
self.0.background.spawn(wait_finished_fut.map(|_| Ok(())));
}
}

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