feat(ecstore): attribute manual transition worker failures (#5324)

* feat(ecstore): attribute manual transition worker failures

Add manual transition worker failure reason tracking and persistence recovery compatibility for checksum validation.

Co-Authored-By: heihutu <heihutu@gmail.com>

* chore(ilm): add manual transition diagnostics scripts

Co-Authored-By: heihutu <heihutu@gmail.com>

* style(ecstore): format manual transition attribution

Co-Authored-By: heihutu <heihutu@gmail.com>

* fix(ecstore): avoid copying failure reasons via clone

Co-Authored-By: heihutu <heihutu@gmail.com>

---------

Co-authored-by: heihutu <heihutu@gmail.com>
This commit is contained in:
houseme
2026-07-27 14:06:12 +08:00
committed by GitHub
parent f566b382a0
commit 9d84056d7b
8 changed files with 1887 additions and 56 deletions
Generated
+23 -23
View File
@@ -56,9 +56,9 @@ dependencies = [
[[package]]
name = "aes"
version = "0.9.1"
version = "0.9.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f1fc76eaeac4c9164506c466d4ffdd8ec9d0c5bf57ee97177c4d8eceb3a0e138"
checksum = "f8eb277bec05f56a0e0591f155a484cbd0f4f07ff2905051a48c72f004f7ed58"
dependencies = [
"cipher 0.5.2",
"cpubits",
@@ -73,7 +73,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fdf011db2e21ce0d575593d749db5554b47fed37aff429e4dc50bc91ac93a028"
dependencies = [
"aead",
"aes 0.9.1",
"aes 0.9.2",
"cipher 0.5.2",
"ctr",
"ghash",
@@ -1528,7 +1528,7 @@ version = "0.10.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3078c7629b62d3f0439517fa394996acacc5cbc91c5a20d8c658e77abd503a71"
dependencies = [
"generic-array 0.14.7",
"generic-array 0.14.9",
]
[[package]]
@@ -1547,7 +1547,7 @@ version = "0.3.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a8894febbff9f758034a5b8e12d87918f56dfc64a8e1fe757d65e29041538d93"
dependencies = [
"generic-array 0.14.7",
"generic-array 0.14.9",
]
[[package]]
@@ -1870,7 +1870,7 @@ version = "0.4.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "773f3b9af64447d2ce9850330c473515014aa235e6a783b02db81ff39e4a3dad"
dependencies = [
"crypto-common 0.1.7",
"crypto-common 0.1.6",
"inout 0.1.4",
]
@@ -2328,7 +2328,7 @@ version = "0.5.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0dc92fb57ca44df6db8059111ab3af99a63d5d0f8375d9972e319a379c6bab76"
dependencies = [
"generic-array 0.14.7",
"generic-array 0.14.9",
"rand_core 0.6.4",
"subtle",
"zeroize",
@@ -2353,11 +2353,11 @@ dependencies = [
[[package]]
name = "crypto-common"
version = "0.1.7"
version = "0.1.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "78c8292055d1c1df0cce5d180393dc8cce0abec0a7102adb6c7b1eef6016d60a"
checksum = "1bfb12502f3fc46cca1bb51ac28df9d618d813cdc3d2f25b9fe775a34af26bb3"
dependencies = [
"generic-array 0.14.7",
"generic-array 0.14.9",
"typenum",
]
@@ -3554,7 +3554,7 @@ checksum = "9ed9a281f7bc9b7576e61468ba615a66a5c8cfdff42420a70aa82701a3b1e292"
dependencies = [
"block-buffer 0.10.4",
"const-oid 0.9.6",
"crypto-common 0.1.7",
"crypto-common 0.1.6",
"subtle",
]
@@ -3811,7 +3811,7 @@ dependencies = [
"crypto-bigint 0.5.5",
"digest 0.10.7",
"ff 0.13.1",
"generic-array 0.14.7",
"generic-array 0.14.9",
"group 0.13.0",
"hkdf 0.12.4",
"pem-rfc7468 0.7.0",
@@ -4257,9 +4257,9 @@ dependencies = [
[[package]]
name = "generic-array"
version = "0.14.7"
version = "0.14.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "85649ca51fd72272d7821adaf274ad91c288277713d9c18820d8499a7ff69e9a"
checksum = "4bb6743198531e02858aeaea5398fcc883e71851fcbcb5a2f773e2fb6cb1edf2"
dependencies = [
"typenum",
"version_check",
@@ -4272,7 +4272,7 @@ version = "1.4.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ab4e5aa225bc56696909483320f0ff9b600f1a971b52e07a17d70f3d9b43254b"
dependencies = [
"generic-array 0.14.7",
"generic-array 0.14.9",
"rustversion",
"typenum",
]
@@ -5272,7 +5272,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "879f10e63c20629ecabbb64a8010319738c66a5cd0c29b02d63d272b03751d01"
dependencies = [
"block-padding 0.3.3",
"generic-array 0.14.7",
"generic-array 0.14.9",
]
[[package]]
@@ -5745,9 +5745,9 @@ checksum = "3eaf3ede3fee6db1a4c2ee091bf8a8b4dccdc6d17f656fb07896ee72867612f2"
[[package]]
name = "libflate"
version = "2.3.0"
version = "2.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cd96e993e5f3368b0cb8497dae6c860c22af8ff18388c61c6c0b86c58d86b5df"
checksum = "a4da9b700e758e57152a1fd1c52cbdc5727c1aa6d8743dc1acda917398f1d76c"
dependencies = [
"adler32",
"crc32fast",
@@ -7485,7 +7485,7 @@ version = "0.8.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "63d440a804ec8d6fafbb6b84471e013286658d373248927692ab3366686220ca"
dependencies = [
"aes 0.9.1",
"aes 0.9.2",
"aes-gcm",
"cbc 0.2.1",
"der 0.8.1",
@@ -8681,7 +8681,7 @@ version = "0.62.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b8b67b5a0d8068c89dcbe9d95df986af7a851d1f3c604525274c37468e60464f"
dependencies = [
"aes 0.9.1",
"aes 0.9.2",
"aws-lc-rs",
"bitflags 2.13.1",
"block-padding 0.4.2",
@@ -10503,7 +10503,7 @@ checksum = "d3e97a565f76233a6003f9f5c54be1d9c5bdfa3eccfb189469f11ec4901c47dc"
dependencies = [
"base16ct 0.2.0",
"der 0.7.10",
"generic-array 0.14.7",
"generic-array 0.14.9",
"pkcs8 0.10.2",
"subtle",
"zeroize",
@@ -11123,7 +11123,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d801accda99469cde6d73da741422610fdf6508a72d9a69d1b55cb241c720597"
dependencies = [
"aead",
"aes 0.9.1",
"aes 0.9.2",
"aes-gcm",
"chacha20",
"cipher 0.5.2",
@@ -13068,7 +13068,7 @@ version = "8.6.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2d04a6b5381502aa6087c94c669499eb1602eb9c5e8198e534de571f7154809b"
dependencies = [
"aes 0.9.1",
"aes 0.9.2",
"bzip2",
"constant_time_eq",
"crc32fast",
+3 -2
View File
@@ -68,7 +68,7 @@ resolver = "3"
edition = "2024"
license = "Apache-2.0"
repository = "https://github.com/rustfs/rustfs"
rust-version = "1.96.0"
rust-version = "1.97.1"
version = "1.0.0-beta.11"
homepage = "https://rustfs.com"
description = "RustFS is a high-performance distributed object storage software built using Rust, one of the most popular languages worldwide. "
@@ -211,7 +211,7 @@ zeroize = { version = "1.9.0" }
# Time and Date
chrono = { version = "0.4.45" }
humantime = "2.4.0"
jiff = { version = "0.2.34" }
jiff = { version = "0.2.35" }
time = { version = "0.3.54" }
# Database
@@ -243,6 +243,7 @@ crossbeam-channel = "0.5.16"
crossbeam-deque = "0.8.7"
crossbeam-utils = "0.8.22"
datafusion = { default-features = false, git = "https://github.com/apache/datafusion.git", rev = "dae03ee062b2abf986de8df12ea82fb1578a2d99" }
#datafusion = { default-features = false, version = "54.1.0" }
derive_builder = "0.20.2"
enumset = "1.1.14"
faster-hex = "0.10.0"
@@ -23,13 +23,13 @@ use crate::bucket::lifecycle::lifecycle::{
};
use crate::bucket::lifecycle::manual_transition_job::{
MANUAL_TRANSITION_JOB_RECORD_PREFIX, ManualTransitionJobRecord, ManualTransitionJobState, ManualTransitionScopeAdmission,
ManualTransitionScopeAdmissionClaim, ManualTransitionTaskRecord, ManualTransitionWorkerResult,
claim_manual_transition_scope_admission, delete_manual_transition_scope_admission_if_current,
ManualTransitionScopeAdmissionClaim, ManualTransitionTaskRecord, ManualTransitionWorkerFailureReason,
ManualTransitionWorkerResult, claim_manual_transition_scope_admission, delete_manual_transition_scope_admission_if_current,
load_manual_transition_job_record, load_manual_transition_job_record_with_etag,
manual_transition_job_id_from_record_object_name, manual_transition_job_lease_expired,
manual_transition_worker_result_task_key, persist_manual_transition_job_progress, reconcile_manual_transition_worker_results,
record_manual_transition_worker_result, renew_manual_transition_job_lease, save_manual_transition_job_record_if_current,
save_manual_transition_task_if_absent,
record_manual_transition_worker_result, record_manual_transition_worker_result_with_reason,
renew_manual_transition_job_lease, save_manual_transition_job_record_if_current, save_manual_transition_task_if_absent,
};
use crate::bucket::lifecycle::replication_sink;
use crate::bucket::lifecycle::replication_sink::{
@@ -49,7 +49,9 @@ use crate::disk::error::DiskError;
use crate::disk::{DeleteOptions, Disk, DiskAPI, RUSTFS_META_BUCKET, RUSTFS_META_MULTIPART_BUCKET, STORAGE_FORMAT_FILE};
use crate::error::Error;
use crate::error::StorageError;
use crate::error::{error_resp_to_object_err, is_err_object_not_found, is_err_version_not_found, is_network_or_host_down};
use crate::error::{
error_resp_to_object_err, is_err_object_not_found, is_err_read_quorum, is_err_version_not_found, is_network_or_host_down,
};
use crate::object_api::{GetObjectReader, ObjectInfo, ObjectOptions};
use crate::services::tier::{
tier::{TierConfigMgr, TierOperationLease, tier_destination_id_from_metadata},
@@ -94,7 +96,7 @@ use s3s::dto::{
use s3s::header::{X_AMZ_RESTORE, X_AMZ_SERVER_SIDE_ENCRYPTION};
use sha2::{Digest, Sha256};
use std::any::Any;
use std::collections::{HashMap, HashSet};
use std::collections::{BTreeMap, HashMap, HashSet};
use std::env;
use std::future::Future;
use std::pin::Pin;
@@ -1658,11 +1660,12 @@ impl TransitionState {
.await
{
if let (Some(job_id), Some(result_key)) = (task.manual_job_id, task.manual_result_key.as_deref()) {
record_manual_transition_worker_result_for_task(
record_manual_transition_worker_result_for_task_with_reason(
api.clone(),
job_id,
result_key,
ManualTransitionWorkerResult::TierFailure,
manual_transition_worker_failure_reason(&err),
)
.await;
}
@@ -1818,6 +1821,78 @@ async fn record_manual_transition_worker_result_for_task(
}
}
async fn record_manual_transition_worker_result_for_task_with_reason(
api: Arc<ECStore>,
job_id: Uuid,
result_key: &str,
result: ManualTransitionWorkerResult,
reason: ManualTransitionWorkerFailureReason,
) {
if let Err(err) = record_manual_transition_worker_result_with_reason(
api,
job_id,
result_key,
result,
manual_transition_queue_snapshot(),
Some(reason),
)
.await
{
warn!(
event = EVENT_LIFECYCLE_WORKER_STATE,
component = LOG_COMPONENT_ECSTORE,
subsystem = LOG_SUBSYSTEM_LIFECYCLE,
job_id = %job_id,
error = %err,
reason = ?reason,
state = "manual_transition_worker_result_failed",
"Manual transition worker failed to persist job result"
);
}
}
fn manual_transition_worker_failure_reason(err: &Error) -> ManualTransitionWorkerFailureReason {
if is_err_object_not_found(err) || is_err_version_not_found(err) {
return ManualTransitionWorkerFailureReason::NotFound;
}
if is_err_permission_denied(err) {
return ManualTransitionWorkerFailureReason::PermissionDenied;
}
if is_err_read_quorum(err) {
return ManualTransitionWorkerFailureReason::Quorum;
}
if is_timeout(err) {
return ManualTransitionWorkerFailureReason::Timeout;
}
if is_slow_down(err) {
return ManualTransitionWorkerFailureReason::SlowDown;
}
if is_network_or_host_down(&err.to_string(), false) {
return ManualTransitionWorkerFailureReason::Network;
}
ManualTransitionWorkerFailureReason::Unknown
}
fn is_err_permission_denied(err: &Error) -> bool {
match err {
Error::VolumeAccessDenied | Error::FileAccessDenied | Error::PrefixAccessDenied(_, _) | Error::DiskAccessDenied => true,
Error::Io(io) => io.kind() == std::io::ErrorKind::PermissionDenied,
_ => false,
}
}
fn is_timeout(err: &Error) -> bool {
match err {
Error::Timeout => true,
Error::Io(io) => io.kind() == std::io::ErrorKind::TimedOut,
_ => false,
}
}
fn is_slow_down(err: &Error) -> bool {
matches!(err, Error::SlowDown)
}
pub async fn init_background_expiry(api: Arc<ECStore>) {
let mut workers = get_env_usize("RUSTFS_MAX_EXPIRY_WORKERS", std::cmp::min(num_cpus::get(), 16));
//globalILMConfig.getExpirationWorkers()
@@ -3233,6 +3308,8 @@ pub struct ManualTransitionRunReport {
#[serde(default, skip_serializing_if = "is_zero_u64")]
pub transition_failed: u64,
pub tier_failure: u64,
#[serde(default, skip_serializing_if = "BTreeMap::is_empty")]
pub tier_failure_by_reason: BTreeMap<ManualTransitionWorkerFailureReason, u64>,
pub truncated_by_limit: bool,
pub truncated_by_duration: bool,
pub cancelled: bool,
@@ -3303,12 +3380,18 @@ impl ManualTransitionRunReport {
}
pub fn merge_scan_report_preserving_worker(&mut self, scan_report: &ManualTransitionRunReport) {
let mut tier_failure_by_reason = self.tier_failure_by_reason.clone();
for (reason, count) in &scan_report.tier_failure_by_reason {
let current = tier_failure_by_reason.get(reason).copied().unwrap_or_default();
tier_failure_by_reason.insert(*reason, current.max(*count));
}
let transition_completed = self.transition_completed;
let transition_failed = self.transition_failed;
*self = scan_report.clone();
self.transition_completed = transition_completed;
self.transition_failed = transition_failed;
self.tier_failure = scan_report.tier_failure.saturating_add(transition_failed);
self.tier_failure_by_reason = tier_failure_by_reason;
}
pub fn worker_transition_pending(&self) -> bool {
@@ -4658,12 +4741,13 @@ mod tests {
lifecycle_delete_all_versions_replication_scan, lifecycle_deleted_object, lifecycle_replication_blocks_action,
lifecycle_rule_has_date_expiration, lifecycle_version_purge_state_from_completed_targets,
manual_transition_duration_elapsed, manual_transition_has_more_after_limit, manual_transition_recovery_progress_sink,
manual_transition_version_marker, mark_delete_opts_skip_decommissioned_on_remote_success,
merge_stale_multipart_candidate, persist_manual_transition_job_progress, persist_manual_transition_page_checkpoint,
recover_manual_transition_job, recover_manual_transition_jobs, replication_state_for_delete,
resolve_tier_free_version_recovery_enabled, resolve_transition_queue_capacity, resolve_transition_queue_send_timeout,
resolve_transition_worker_count, resolve_transition_workers_absolute_max, run_tier_free_version_recovery_loop,
select_restore_s3_location, set_lifecycle_observability_observer, set_recovered_free_version_enqueue_observer,
manual_transition_version_marker, manual_transition_worker_failure_reason,
mark_delete_opts_skip_decommissioned_on_remote_success, merge_stale_multipart_candidate,
persist_manual_transition_job_progress, persist_manual_transition_page_checkpoint, recover_manual_transition_job,
recover_manual_transition_jobs, replication_state_for_delete, resolve_tier_free_version_recovery_enabled,
resolve_transition_queue_capacity, resolve_transition_queue_send_timeout, resolve_transition_worker_count,
resolve_transition_workers_absolute_max, run_tier_free_version_recovery_loop, select_restore_s3_location,
set_lifecycle_observability_observer, set_recovered_free_version_enqueue_observer,
should_defer_date_expiry_for_recent_config_update, should_reuse_lifecycle_delete_replication_state,
transitioned_cleanup_tuple, transitioned_object_delete_opts, wait_for_tier_free_version_recovery,
};
@@ -4676,16 +4760,17 @@ mod tests {
use crate::bucket::lifecycle::config_boundary;
use crate::bucket::lifecycle::manual_transition_job::{
ManualTransitionJobRecord, ManualTransitionJobState, ManualTransitionScopeAdmission, ManualTransitionScopeAdmissionClaim,
ManualTransitionTaskRecord, ManualTransitionWorkerResult, ManualTransitionWorkerResultRecord,
claim_manual_transition_scope_admission, delete_manual_transition_scope_admission_if_current,
legacy_manual_transition_scope_key, load_manual_transition_job_record, load_manual_transition_scope_admission,
ManualTransitionTaskRecord, ManualTransitionWorkerFailureReason, ManualTransitionWorkerResult,
ManualTransitionWorkerResultRecord, claim_manual_transition_scope_admission,
delete_manual_transition_scope_admission_if_current, legacy_manual_transition_scope_key,
load_manual_transition_job_record, load_manual_transition_scope_admission,
load_manual_transition_scope_admission_with_etag, load_manual_transition_task_record,
manual_transition_scope_record_object_name, manual_transition_worker_result_object_name,
manual_transition_worker_result_task_key, reconcile_manual_transition_worker_results,
record_manual_transition_worker_result, renew_manual_transition_job_lease, request_manual_transition_job_cancel,
save_manual_transition_job_record, save_manual_transition_scope_admission_if_absent,
save_manual_transition_scope_admission_if_current, save_manual_transition_task_if_absent,
save_manual_transition_worker_result_if_absent,
record_manual_transition_worker_result, record_manual_transition_worker_result_with_reason,
renew_manual_transition_job_lease, request_manual_transition_job_cancel, save_manual_transition_job_record,
save_manual_transition_scope_admission_if_absent, save_manual_transition_scope_admission_if_current,
save_manual_transition_task_if_absent, save_manual_transition_worker_result_if_absent,
};
use crate::bucket::lifecycle::replication_sink::{
ReplicateDecision, ReplicateTargetDecision, ReplicationStatusType, VersionPurgeStatusType,
@@ -7539,6 +7624,37 @@ mod tests {
assert!(record.completed_at_unix_nanos.is_some());
}
#[test]
fn manual_transition_worker_failure_reason_classifies_known_failures() {
assert_eq!(
manual_transition_worker_failure_reason(&Error::ObjectNotFound("bucket".to_string(), "obj".to_string())),
ManualTransitionWorkerFailureReason::NotFound
);
assert_eq!(
manual_transition_worker_failure_reason(&Error::Io(std::io::Error::new(
std::io::ErrorKind::PermissionDenied,
"file access denied",
))),
ManualTransitionWorkerFailureReason::PermissionDenied
);
assert_eq!(
manual_transition_worker_failure_reason(&Error::ErasureReadQuorum),
ManualTransitionWorkerFailureReason::Quorum
);
assert_eq!(
manual_transition_worker_failure_reason(&Error::Timeout),
ManualTransitionWorkerFailureReason::Timeout
);
assert_eq!(
manual_transition_worker_failure_reason(&Error::SlowDown),
ManualTransitionWorkerFailureReason::SlowDown
);
assert_eq!(
manual_transition_worker_failure_reason(&Error::MethodNotAllowed),
ManualTransitionWorkerFailureReason::Unknown
);
}
#[tokio::test]
async fn manual_transition_counts_already_transitioned_object() {
let lc = latest_transition_lifecycle();
@@ -8471,6 +8587,49 @@ mod tests {
);
}
#[tokio::test]
#[serial]
async fn manual_transition_worker_result_stores_failure_reason() {
let (_paths, ecstore) = setup_test_env().await;
let job_id = Uuid::new_v4();
let bucket = format!("manual-worker-failure-reason-{}", job_id.simple());
let options = ManualTransitionRunOptions::default();
let mut record = ManualTransitionJobRecord::new(job_id, &bucket, &options, "worker-owner");
record.complete(
ManualTransitionRunReport {
bucket: bucket.clone(),
enqueued: 1,
..Default::default()
},
ManualTransitionQueueSnapshot::default(),
);
save_manual_transition_job_record(ecstore.clone(), &record)
.await
.expect("worker result job record should save");
let task_key = manual_transition_worker_result_task_key(&bucket, "logs/fail", None);
let final_record = record_manual_transition_worker_result_with_reason(
ecstore.clone(),
job_id,
&task_key,
ManualTransitionWorkerResult::TierFailure,
ManualTransitionQueueSnapshot::default(),
Some(ManualTransitionWorkerFailureReason::Network),
)
.await
.expect("worker result with failure reason should persist");
assert_eq!(
final_record
.report
.tier_failure_by_reason
.get(&ManualTransitionWorkerFailureReason::Network),
Some(&1)
);
assert_eq!(final_record.report.transition_failed, 1);
assert_eq!(final_record.report.tier_failure, 1);
}
#[tokio::test]
#[serial]
async fn manual_transition_worker_result_reconcile_applies_marker_and_releases_admission() {
@@ -12,6 +12,7 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use std::collections::BTreeMap;
use std::sync::Arc;
use rustfs_utils::crypto::{hex_sha256, is_sha256_checksum};
@@ -194,6 +195,15 @@ impl ManualTransitionJobRecord {
}
pub fn record_worker_result(&mut self, result: ManualTransitionWorkerResult, queue_snapshot: ManualTransitionQueueSnapshot) {
self.record_worker_result_with_reason(result, queue_snapshot, None);
}
pub fn record_worker_result_with_reason(
&mut self,
result: ManualTransitionWorkerResult,
queue_snapshot: ManualTransitionQueueSnapshot,
failure_reason: Option<ManualTransitionWorkerFailureReason>,
) {
if self.is_terminal() {
return;
}
@@ -204,6 +214,8 @@ impl ManualTransitionJobRecord {
ManualTransitionWorkerResult::TierFailure => {
self.report.transition_failed = self.report.transition_failed.saturating_add(1);
self.report.tier_failure = self.report.tier_failure.saturating_add(1);
let reason = failure_reason.unwrap_or(ManualTransitionWorkerFailureReason::Unknown);
*self.report.tier_failure_by_reason.entry(reason).or_insert(0) += 1;
}
}
self.queue_snapshot = queue_snapshot;
@@ -215,6 +227,7 @@ impl ManualTransitionJobRecord {
&mut self,
completed: u64,
failed: u64,
failure_reasons: &BTreeMap<ManualTransitionWorkerFailureReason, u64>,
task_queued: u64,
queue_snapshot: ManualTransitionQueueSnapshot,
) -> bool {
@@ -237,11 +250,17 @@ impl ManualTransitionJobRecord {
{
return false;
}
let mut scan_tier_failure_by_reason = self.report.tier_failure_by_reason.clone();
for (reason, count) in failure_reasons {
let current = scan_tier_failure_by_reason.get(reason).copied().unwrap_or_default();
scan_tier_failure_by_reason.insert(*reason, current.max(*count));
}
let scan_tier_failure = self.report.tier_failure.saturating_sub(self.report.transition_failed);
self.report.enqueued = enqueued;
self.report.transition_completed = transition_completed;
self.report.transition_failed = transition_failed;
self.report.tier_failure = scan_tier_failure.saturating_add(transition_failed);
self.report.tier_failure_by_reason = scan_tier_failure_by_reason;
self.queue_snapshot = queue_snapshot;
self.updated_at_unix_nanos = OffsetDateTime::now_utc().unix_timestamp_nanos();
self.mark_terminal_if_worker_drained();
@@ -401,9 +420,24 @@ impl ManualTransitionJobRecord {
return Err(ManualTransitionJobError::Corrupt("content checksum is not a sha256 checksum"));
}
let mut job = persisted.job;
let job_bytes = serde_json::to_vec(&job)?;
let mut job_bytes = serde_json::to_vec(&job)?;
let actual_checksum = hex_sha256(&job_bytes, ToOwned::to_owned);
if persisted.content_sha256 != actual_checksum {
let checksum_match = if persisted.content_sha256 == actual_checksum {
true
} else {
let mut job_value = serde_json::to_value(&job)?;
if let Some(job_obj) = job_value.as_object_mut() {
if let Some(report) = job_obj.get_mut("report").and_then(|value| value.as_object_mut()) {
report.remove("tier_failure_by_reason");
}
} else {
return Err(ManualTransitionJobError::ChecksumMismatch);
}
job_bytes = serde_json::to_vec(&job_value)?;
let fallback_checksum = hex_sha256(&job_bytes, ToOwned::to_owned);
persisted.content_sha256 == fallback_checksum
};
if !checksum_match {
return Err(ManualTransitionJobError::ChecksumMismatch);
}
if job.job_id != expected_job_id {
@@ -449,6 +483,18 @@ pub enum ManualTransitionWorkerResult {
TierFailure,
}
#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Serialize, Deserialize)]
#[serde(rename_all = "snake_case")]
pub enum ManualTransitionWorkerFailureReason {
Unknown,
NotFound,
Network,
PermissionDenied,
Timeout,
Quorum,
SlowDown,
}
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
#[serde(deny_unknown_fields)]
pub struct ManualTransitionTaskRecord {
@@ -554,17 +600,22 @@ struct PersistedManualTransitionTaskRecord {
record: ManualTransitionTaskRecord,
}
#[derive(Debug, Clone, Copy, Default, PartialEq, Eq)]
#[derive(Debug, Clone, Default, PartialEq, Eq)]
pub struct ManualTransitionWorkerResultStats {
pub completed: u64,
pub failed: u64,
pub tier_failure_by_reason: BTreeMap<ManualTransitionWorkerFailureReason, u64>,
}
impl ManualTransitionWorkerResultStats {
fn record(&mut self, result: ManualTransitionWorkerResult) {
fn record(&mut self, result: ManualTransitionWorkerResult, failure_reason: Option<ManualTransitionWorkerFailureReason>) {
match result {
ManualTransitionWorkerResult::Completed => self.completed = self.completed.saturating_add(1),
ManualTransitionWorkerResult::TierFailure => self.failed = self.failed.saturating_add(1),
ManualTransitionWorkerResult::TierFailure => {
self.failed = self.failed.saturating_add(1);
let reason = failure_reason.unwrap_or(ManualTransitionWorkerFailureReason::Unknown);
*self.tier_failure_by_reason.entry(reason).or_insert(0) += 1;
}
}
}
}
@@ -576,16 +627,28 @@ pub struct ManualTransitionWorkerResultRecord {
pub job_id: Uuid,
pub task_key: String,
pub result: ManualTransitionWorkerResult,
#[serde(default, skip_serializing_if = "Option::is_none")]
pub failure_reason: Option<ManualTransitionWorkerFailureReason>,
pub completed_at_unix_nanos: i128,
}
impl ManualTransitionWorkerResultRecord {
pub fn new(job_id: Uuid, task_key: impl Into<String>, result: ManualTransitionWorkerResult) -> Self {
Self::new_with_reason(job_id, task_key, result, None)
}
pub fn new_with_reason(
job_id: Uuid,
task_key: impl Into<String>,
result: ManualTransitionWorkerResult,
failure_reason: Option<ManualTransitionWorkerFailureReason>,
) -> Self {
Self {
schema: MANUAL_TRANSITION_WORKER_RESULT_SCHEMA.to_string(),
job_id,
task_key: task_key.into(),
result,
failure_reason,
completed_at_unix_nanos: OffsetDateTime::now_utc().unix_timestamp_nanos(),
}
}
@@ -620,9 +683,23 @@ impl ManualTransitionWorkerResultRecord {
));
}
let record = persisted.record;
let record_bytes = serde_json::to_vec(&record)?;
let mut record_bytes = serde_json::to_vec(&record)?;
let actual_checksum = hex_sha256(&record_bytes, ToOwned::to_owned);
if persisted.content_sha256 != actual_checksum {
let checksum_match = if persisted.content_sha256 == actual_checksum {
true
} else {
if let Some(record_value) = serde_json::to_value(&record)?.as_object_mut() {
if record.failure_reason.is_none() {
record_value.remove("failure_reason");
}
record_bytes = serde_json::to_vec(record_value)?;
} else {
return Err(ManualTransitionJobError::ChecksumMismatch);
}
let fallback_checksum = hex_sha256(&record_bytes, ToOwned::to_owned);
persisted.content_sha256 == fallback_checksum
};
if !checksum_match {
return Err(ManualTransitionJobError::ChecksumMismatch);
}
if record.job_id != expected_job_id {
@@ -1120,7 +1197,7 @@ async fn scan_manual_transition_worker_result_journal(
Ok(result) => result,
Err(err) => return Ok(ManualTransitionWorkerResultJournal::Corrupt(err.to_string())),
};
stats.record(result.result);
stats.record(result.result, result.failure_reason);
}
if !page.is_truncated {
return Ok(ManualTransitionWorkerResultJournal::Stats(stats));
@@ -1151,7 +1228,13 @@ pub async fn reconcile_manual_transition_worker_results(
};
for _ in 0..4 {
let (mut record, etag) = load_manual_transition_job_record_with_etag(api.clone(), job_id).await?;
let changed = record.apply_worker_result_counts(stats.completed, stats.failed, task_stats.queued, queue_snapshot);
let changed = record.apply_worker_result_counts(
stats.completed,
stats.failed,
&stats.tier_failure_by_reason,
task_stats.queued,
queue_snapshot,
);
if !changed {
return Ok(record);
}
@@ -1446,7 +1529,18 @@ pub async fn record_manual_transition_worker_result(
result: ManualTransitionWorkerResult,
queue_snapshot: ManualTransitionQueueSnapshot,
) -> EcstoreResult<ManualTransitionJobRecord> {
let result_record = ManualTransitionWorkerResultRecord::new(job_id, task_key, result);
record_manual_transition_worker_result_with_reason(api, job_id, task_key, result, queue_snapshot, None).await
}
pub async fn record_manual_transition_worker_result_with_reason(
api: Arc<ECStore>,
job_id: Uuid,
task_key: &str,
result: ManualTransitionWorkerResult,
queue_snapshot: ManualTransitionQueueSnapshot,
failure_reason: Option<ManualTransitionWorkerFailureReason>,
) -> EcstoreResult<ManualTransitionJobRecord> {
let result_record = ManualTransitionWorkerResultRecord::new_with_reason(job_id, task_key, result, failure_reason);
if !save_manual_transition_worker_result_if_absent(api.clone(), &result_record).await? {
return load_manual_transition_job_record(api, job_id).await;
}
@@ -1456,7 +1550,7 @@ pub async fn record_manual_transition_worker_result(
if record.is_terminal() {
return Ok(record);
}
record.record_worker_result(result, queue_snapshot);
record.record_worker_result_with_reason(result, queue_snapshot, failure_reason);
match save_manual_transition_job_record_if_current(api.clone(), &record, &etag).await {
Ok(()) => {
if record.is_terminal() {
@@ -1652,6 +1746,52 @@ mod tests {
assert_eq!(record.report.tier_failure, 1);
}
#[test]
fn manual_transition_job_record_records_worker_failure_reasons() {
let options = ManualTransitionRunOptions::default();
let mut record = ManualTransitionJobRecord::new(Uuid::new_v4(), "bucket", &options, TEST_OWNER);
record.record_worker_result_with_reason(
ManualTransitionWorkerResult::TierFailure,
ManualTransitionQueueSnapshot::default(),
Some(ManualTransitionWorkerFailureReason::NotFound),
);
record.record_worker_result_with_reason(
ManualTransitionWorkerResult::TierFailure,
ManualTransitionQueueSnapshot::default(),
Some(ManualTransitionWorkerFailureReason::Network),
);
record.record_worker_result_with_reason(
ManualTransitionWorkerResult::TierFailure,
ManualTransitionQueueSnapshot::default(),
None,
);
assert_eq!(record.report.transition_failed, 3);
assert_eq!(record.report.tier_failure, 3);
assert_eq!(
record
.report
.tier_failure_by_reason
.get(&ManualTransitionWorkerFailureReason::NotFound),
Some(&1)
);
assert_eq!(
record
.report
.tier_failure_by_reason
.get(&ManualTransitionWorkerFailureReason::Network),
Some(&1)
);
assert_eq!(
record
.report
.tier_failure_by_reason
.get(&ManualTransitionWorkerFailureReason::Unknown),
Some(&1)
);
}
#[test]
fn manual_transition_job_cancel_recovery_finishes_after_worker_drain() {
let options = ManualTransitionRunOptions::default();
@@ -1709,6 +1849,67 @@ mod tests {
assert_eq!(record.report.tier_failure, 3);
}
#[test]
fn manual_transition_job_scan_progress_preserves_worker_failure_reasons() {
let options = ManualTransitionRunOptions::default();
let mut record = ManualTransitionJobRecord::new(Uuid::new_v4(), "bucket", &options, TEST_OWNER);
record.record_worker_result_with_reason(
ManualTransitionWorkerResult::TierFailure,
ManualTransitionQueueSnapshot::default(),
Some(ManualTransitionWorkerFailureReason::NotFound),
);
record.update_running_progress(
ManualTransitionRunReport {
bucket: "bucket".to_string(),
scanned: 3,
enqueued: 1,
tier_failure: 2,
..Default::default()
},
ManualTransitionQueueSnapshot::default(),
);
assert_eq!(
record
.report
.tier_failure_by_reason
.get(&ManualTransitionWorkerFailureReason::NotFound),
Some(&1)
);
}
#[test]
fn manual_transition_job_apply_worker_result_counts_preserves_existing_failure_reasons() {
let options = ManualTransitionRunOptions::default();
let mut record = ManualTransitionJobRecord::new(Uuid::new_v4(), "bucket", &options, TEST_OWNER);
record.record_worker_result_with_reason(
ManualTransitionWorkerResult::TierFailure,
ManualTransitionQueueSnapshot::default(),
Some(ManualTransitionWorkerFailureReason::NotFound),
);
let mut failure_reasons = BTreeMap::new();
failure_reasons.insert(ManualTransitionWorkerFailureReason::PermissionDenied, 2);
record.apply_worker_result_counts(0, 1, &failure_reasons, 1, ManualTransitionQueueSnapshot::default());
assert_eq!(
record
.report
.tier_failure_by_reason
.get(&ManualTransitionWorkerFailureReason::NotFound),
Some(&1)
);
assert_eq!(
record
.report
.tier_failure_by_reason
.get(&ManualTransitionWorkerFailureReason::PermissionDenied),
Some(&2)
);
}
#[test]
fn manual_transition_job_unknown_checkpoint_persists_counters_and_cursor() {
let options = ManualTransitionRunOptions {
@@ -1955,6 +2156,62 @@ mod tests {
assert!(matches!(err, ManualTransitionJobError::ChecksumMismatch));
}
#[test]
fn manual_transition_worker_result_record_round_trips_without_failure_reason() {
let job_id = Uuid::new_v4();
let task_key = manual_transition_worker_result_task_key("bucket", "logs/a", None);
let record = ManualTransitionWorkerResultRecord::new_with_reason(
job_id,
&task_key,
ManualTransitionWorkerResult::TierFailure,
Some(ManualTransitionWorkerFailureReason::Network),
);
let encoded = record.encode().expect("worker result record should encode");
let mut value: serde_json::Value = serde_json::from_slice(&encoded).expect("encoded worker result should be json");
let record_obj = value["record"].as_object_mut().expect("record object should be present");
record_obj.remove("failure_reason");
let record_bytes = serde_json::to_vec(&value["record"]).expect("record without reason should encode");
value["content_sha256"] = serde_json::Value::String(hex_sha256(&record_bytes, ToOwned::to_owned));
let stripped = serde_json::to_vec(&value).expect("stripped worker result should encode");
let decoded = ManualTransitionWorkerResultRecord::decode(job_id, &task_key, &stripped)
.expect("worker result without reason should decode");
assert_eq!(decoded.failure_reason, None);
assert_eq!(decoded.result, ManualTransitionWorkerResult::TierFailure);
}
#[test]
fn manual_transition_worker_result_stats_aggregates_reasons() {
let mut stats = ManualTransitionWorkerResultStats::default();
stats.record(ManualTransitionWorkerResult::Completed, None);
stats.record(
ManualTransitionWorkerResult::TierFailure,
Some(ManualTransitionWorkerFailureReason::PermissionDenied),
);
stats.record(
ManualTransitionWorkerResult::TierFailure,
Some(ManualTransitionWorkerFailureReason::PermissionDenied),
);
stats.record(ManualTransitionWorkerResult::TierFailure, None);
assert_eq!(stats.completed, 1);
assert_eq!(stats.failed, 3);
assert_eq!(
stats
.tier_failure_by_reason
.get(&ManualTransitionWorkerFailureReason::PermissionDenied),
Some(&2)
);
assert_eq!(
stats
.tier_failure_by_reason
.get(&ManualTransitionWorkerFailureReason::Unknown),
Some(&1)
);
}
#[test]
fn manual_transition_task_record_rejects_checksum_drift() {
let job_id = Uuid::new_v4();
@@ -1971,6 +2228,24 @@ mod tests {
assert!(matches!(err, ManualTransitionJobError::ChecksumMismatch));
}
#[test]
fn manual_transition_job_record_round_trips_legacy_report_without_failure_reason_map() {
let options = ManualTransitionRunOptions::default();
let record = ManualTransitionJobRecord::new(Uuid::new_v4(), "bucket", &options, TEST_OWNER);
let encoded = record.encode().expect("job record should encode");
let mut value: serde_json::Value = serde_json::from_slice(&encoded).expect("encoded job should be json");
let report = value["job"]["report"].as_object_mut().expect("report should be object");
report.remove("tier_failure_by_reason");
let record_bytes = serde_json::to_vec(&value["job"]).expect("job without report failure map should encode");
value["content_sha256"] = serde_json::Value::String(hex_sha256(&record_bytes, ToOwned::to_owned));
let legacy = serde_json::to_vec(&value).expect("job without report failure map should encode");
let decoded = ManualTransitionJobRecord::decode(record.job_id, &legacy).expect("legacy job report should decode");
assert_eq!(decoded.state, ManualTransitionJobState::Running);
assert!(decoded.report.tier_failure_by_reason.is_empty());
}
#[test]
fn manual_transition_job_record_rejects_unknown_report_fields() {
let options = ManualTransitionRunOptions::default();
+253
View File
@@ -0,0 +1,253 @@
#!/usr/bin/env bash
#
# Copyright 2026 RustFS Team
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
set -euo pipefail
ENDPOINT=""
JOB_ID=""
ADMIN_TOKEN=""
LOG_GLOB="/var/log/rustfs/*.log"
METRIC_TYPES="1"
METRIC_SAMPLES="1"
METRIC_INTERVAL=""
BUCKET_FILTER=""
SHOW_HELP=false
usage() {
cat <<'USAGE'
Usage:
scripts/manual_transition_debug.sh --endpoint <admin-api-base> --job-id <job-id> [options]
Required:
--endpoint Admin endpoint base, e.g. http://127.0.0.1:9000
--job-id Manual transition job_id (UUID)
Optional:
--admin-token Bearer token for /admin/v3 endpoints
--log-glob Log glob to scan, defaults to /var/log/rustfs/*.log
--metric-types Metrics types bitmask, defaults to 1 (SCANNER)
--metric-samples Metrics samples, defaults to 1
--metric-interval Metrics interval parameter, passed as-is to query string
--bucket Optional bucket filter for focused log output
-h, --help Show this help
Output sections:
- lifecycle_worker_state structured events (job-scoped)
- manual-transition admin job status
- lifecycle_tier_operation_failed and admin_ilm_transition_state traces
- metrics snapshot (aggregated.scanner.lifecycle_transition)
USAGE
}
require_cmd() {
local cmd="$1"
if ! command -v "$cmd" >/dev/null 2>&1; then
echo "ERROR: required command not found: $cmd" >&2
exit 1
fi
}
api_get() {
local path="$1"
local url="${ENDPOINT%/}${path}"
if [[ -n "$ADMIN_TOKEN" ]]; then
curl -sS -H "Authorization: Bearer ${ADMIN_TOKEN}" "$url"
return
fi
curl -sS "$url"
}
parse_args() {
while [[ $# -gt 0 ]]; do
case "$1" in
--endpoint)
ENDPOINT="${2:-}"
shift 2
;;
--job-id)
JOB_ID="${2:-}"
shift 2
;;
--admin-token)
ADMIN_TOKEN="${2:-}"
shift 2
;;
--log-glob)
LOG_GLOB="${2:-}"
shift 2
;;
--metric-types)
METRIC_TYPES="${2:-}"
shift 2
;;
--metric-samples)
METRIC_SAMPLES="${2:-}"
shift 2
;;
--metric-interval)
METRIC_INTERVAL="${2:-}"
shift 2
;;
--bucket)
BUCKET_FILTER="${2:-}"
shift 2
;;
-h|--help)
SHOW_HELP=true
shift
;;
*)
echo "ERROR: unknown arg: $1" >&2
usage
exit 1
;;
esac
done
}
validate() {
if [[ "$SHOW_HELP" == true ]]; then
usage
exit 0
fi
if [[ -z "$ENDPOINT" || -z "$JOB_ID" ]]; then
echo "ERROR: --endpoint and --job-id are required" >&2
usage
exit 1
fi
if [[ -z "$LOG_GLOB" ]]; then
echo "ERROR: --log-glob must not be empty" >&2
exit 1
fi
}
build_admin_metrics_path() {
local path="/rustfs/admin/v3/metrics?types=${METRIC_TYPES}&n=${METRIC_SAMPLES}"
if [[ -n "$METRIC_INTERVAL" ]]; then
path="${path}&interval=${METRIC_INTERVAL}"
fi
printf '%s' "$path"
}
run_json_stream() {
local query="$1"
local filter="$2"
local log_lines; log_lines="$(mktemp)"
local log_errors; log_errors="$(mktemp)"
local log_paths=()
local match
while IFS= read -r match; do
[[ -n "$match" ]] && log_paths+=("$match")
done < <(compgen -G "$LOG_GLOB" || true)
if [[ "${#log_paths[@]}" -eq 0 ]]; then
log_paths=("$LOG_GLOB")
fi
if ! rg -n --color=never -- "$query" -- "${log_paths[@]}" >"$log_lines" 2>"$log_errors"; then
:
fi
if [[ -s "$log_errors" ]]; then
sed 's/^/DEBUG: /' "$log_errors"
fi
jq -R "$filter" <"$log_lines"
rm -f "$log_lines" "$log_errors"
}
run_worker_logs() {
local base_filter='fromjson? | select(.event == "lifecycle_worker_state")'
local manual_filter='fromjson? | select(.event == "lifecycle_worker_state" and .job_id == env.JOB_ID and (.state | startswith("manual_transition_")))'
local dist_filter='fromjson? | select(.event == "lifecycle_worker_state" and (.state | startswith("manual_transition_")) ) | .state'
if [[ -n "$BUCKET_FILTER" ]]; then
manual_filter='fromjson? | select(.event == "lifecycle_worker_state" and .job_id == env.JOB_ID and (.state | startswith("manual_transition_")) and .bucket == env.BUCKET_FILTER)'
dist_filter='fromjson? | select(.event == "lifecycle_worker_state" and (.state | startswith("manual_transition_")) and .bucket == env.BUCKET_FILTER) | .state'
fi
echo "## lifecycle_worker_state (all)"
run_json_stream "lifecycle_worker_state" "$base_filter"
echo "## lifecycle_worker_state (manual_transition + job filter: ${JOB_ID})"
JOB_ID="$JOB_ID" BUCKET_FILTER="$BUCKET_FILTER" \
run_json_stream "lifecycle_worker_state" "$manual_filter" || true
echo "## lifecycle_worker_state state distribution (manual_transition only)"
JOB_ID="$JOB_ID" BUCKET_FILTER="$BUCKET_FILTER" \
run_json_stream "lifecycle_worker_state" "$dist_filter" | jq -r . | sort | uniq -c | sort -nr || true
}
run_admin_job() {
echo "## manual transition job status"
api_get "/rustfs/admin/v3/ilm/transition/jobs/${JOB_ID}" \
| jq '{status, mode, job_id, failure_reason, report: .report, queue_snapshot: .queue_snapshot}'
}
run_tier_operation_failures() {
echo "## lifecycle_tier_operation_failed events"
local filter='fromjson? | select(.event == "lifecycle_tier_operation_failed") | {timestamp, bucket: .bucket, object: .object, version_id: .version_id, tier: .tier, operation: .operation, error: .error}'
if [[ -n "$BUCKET_FILTER" ]]; then
filter='fromjson? | select(.event == "lifecycle_tier_operation_failed" and .bucket == env.BUCKET_FILTER) | {timestamp, bucket: .bucket, object: .object, version_id: .version_id, tier: .tier, operation: .operation, error: .error}'
fi
JOB_ID="$JOB_ID" BUCKET_FILTER="$BUCKET_FILTER" \
run_json_stream "lifecycle_tier_operation_failed" "$filter" || true
}
run_admin_events() {
echo "## admin_ilm_transition_state (job filter)"
local filter='fromjson? | select(.event == "admin_ilm_transition_state" and .job_id == env.JOB_ID) | {timestamp, operation: .operation, state: .state, result: .result, failure_reason: .failure_reason, bucket: .bucket, prefix: .prefix}'
if [[ -n "$BUCKET_FILTER" ]]; then
filter='fromjson? | select(.event == "admin_ilm_transition_state" and .job_id == env.JOB_ID and .bucket == env.BUCKET_FILTER) | {timestamp, operation: .operation, state: .state, result: .result, failure_reason: .failure_reason, bucket: .bucket, prefix: .prefix}'
fi
JOB_ID="$JOB_ID" BUCKET_FILTER="$BUCKET_FILTER" \
run_json_stream "admin_ilm_transition_state" "$filter" || true
}
run_metrics() {
echo "## metrics: scanner.lifecycle_transition"
local path
path="$(build_admin_metrics_path)"
api_get "$path" | jq '.aggregated.scanner.lifecycle_transition'
echo "## metrics: scanner lifecycle cycle"
api_get "$path" | jq '.aggregated.scanner | {lifecycle_transition: .lifecycle_transition, current_cycle_ilm_actions: .current_cycle_ilm_actions, current_cycle_lifecycle_transition_actions: .current_cycle_lifecycle_transition_actions}'
}
main() {
parse_args "$@"
validate
require_cmd curl
require_cmd jq
require_cmd rg
run_worker_logs
run_admin_events
run_tier_operation_failures
run_admin_job
run_metrics
}
main "$@"
+590
View File
@@ -0,0 +1,590 @@
#!/usr/bin/env bash
# Copyright 2026 RustFS Team
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
PROJECT_ROOT="$(cd "${SCRIPT_DIR}/.." && pwd)"
ENDPOINT=""
JOB_ID=""
ADMIN_TOKEN=""
LOG_GLOB="/var/log/rustfs/*.log"
METRIC_TYPES="1"
METRIC_SAMPLES="1"
METRIC_INTERVAL=""
TASK_BUCKET_FILTER=""
OUT_DIR=""
SYS_BUCKET="${RUSTFS_META_BUCKET:-.rustfs.sys}"
TASK_PREFIX="ilm/manual-transition/tasks"
RESULT_PREFIX="ilm/manual-transition/results"
MC_BIN="${MC_BIN:-mc}"
MC_ALIAS=""
MC_ENDPOINT=""
MC_ACCESS_KEY=""
MC_SECRET_KEY=""
MC_ALIAS_CREATED=false
SHOW_HELP=false
DRY_RUN=false
usage() {
cat <<'USAGE'
Usage:
scripts/manual_transition_journal_audit.sh --endpoint <admin-api> --job-id <job-uuid> [options]
Required:
--endpoint Admin endpoint base, e.g. https://127.0.0.1:9000
--job-id Manual transition job_id
Optional:
--admin-token Bearer token for /admin/v3 endpoints
--log-glob Log glob, default /var/log/rustfs/*.log
--metric-types /admin/v3/metrics type flags, default 1
--metric-samples /admin/v3/metrics sample count, default 1
--metric-interval /admin/v3/metrics interval query param
--bucket-filter Optional bucket filter for logs
--sys-bucket Meta bucket for journal read (default .rustfs.sys)
--mc-alias Reuse existing mc alias (skip alias set)
--mc-endpoint mc endpoint for .rustfs.sys access
--mc-access-key mc access key
--mc-secret-key mc secret key
--out-dir Artifact output dir
--dry-run
--help Show usage
Output:
- Journal CSV/JSON samples for manual transition tasks and worker results
- Reconcile report (task_count/result_count/missing/mismatch)
- /admin/v3 lifecycle_worker_state / admin_ilm_transition_state events
- lifecycle_tier_operation_failed logs
- aggregated lifecycle transition metrics (admin endpoint sample)
USAGE
}
require_bash4() {
if (( BASH_VERSINFO[0] < 4 )); then
echo "ERROR: manual transition journal audit requires bash >= 4.0" >&2
exit 1
fi
}
arg_value() {
local flag="$1"
local value="${2:-}"
if [[ -z "$value" || "$value" == --* ]]; then
echo "ERROR: missing value for $flag" >&2
exit 1
fi
printf '%s' "$value"
}
require_cmd() {
local cmd="$1"
if ! command -v "$cmd" >/dev/null 2>&1; then
echo "ERROR: command not found: $cmd" >&2
exit 1
fi
}
require_nonempty() {
local value="$1"
local label="$2"
if [[ -z "$value" ]]; then
echo "ERROR: $label is required" >&2
exit 1
fi
}
normalize_uuid() {
local raw="${1//-/}"
printf '%s' "${raw,,}"
}
validate_uuid() {
local normalized
normalized="$(normalize_uuid "$1")"
if ! [[ "$normalized" =~ ^[0-9a-f]{32}$ ]]; then
echo "ERROR: invalid uuid (expect UUIDv4): $1" >&2
exit 1
fi
printf '%s' "$normalized"
}
parse_args() {
while [[ $# -gt 0 ]]; do
case "$1" in
--endpoint) ENDPOINT="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--job-id) JOB_ID="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--admin-token) ADMIN_TOKEN="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--log-glob) LOG_GLOB="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--metric-types) METRIC_TYPES="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--metric-samples) METRIC_SAMPLES="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--metric-interval) METRIC_INTERVAL="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--bucket-filter) TASK_BUCKET_FILTER="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--sys-bucket) SYS_BUCKET="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--mc-alias) MC_ALIAS="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--mc-endpoint) MC_ENDPOINT="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--mc-access-key) MC_ACCESS_KEY="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--mc-secret-key) MC_SECRET_KEY="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--out-dir) OUT_DIR="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--dry-run) DRY_RUN=true; shift ;;
-h|--help) SHOW_HELP=true; shift ;;
*)
echo "ERROR: unknown arg: $1" >&2
usage
exit 1
;;
esac
done
}
validate_args() {
if [[ "$SHOW_HELP" == true ]]; then
usage
exit 0
fi
require_nonempty "$ENDPOINT" "--endpoint"
require_nonempty "$JOB_ID" "--job-id"
JOB_ID="$(validate_uuid "$JOB_ID")"
if [[ "$LOG_GLOB" == "" ]]; then
echo "ERROR: --log-glob must not be empty" >&2
exit 1
fi
if [[ -z "$MC_ALIAS" ]]; then
if [[ -z "$MC_ENDPOINT" || -z "$MC_ACCESS_KEY" || -z "$MC_SECRET_KEY" ]]; then
echo "ERROR: either --mc-alias or (--mc-endpoint + --mc-access-key + --mc-secret-key) is required" >&2
exit 1
fi
fi
}
cleanup_mc_alias() {
if [[ "$MC_ALIAS_CREATED" == true ]]; then
"${MC_BIN}" alias remove "$MC_ALIAS" >/dev/null 2>&1 || true
fi
}
setup_output() {
if [[ -z "$OUT_DIR" ]]; then
OUT_DIR="${PROJECT_ROOT}/target/manual-transition-journal-audit/$(date +%Y%m%dT%H%M%S)"
fi
mkdir -p "$OUT_DIR"
SUMMARY_TXT="${OUT_DIR}/audit_summary.txt"
TASKS_CSV="${OUT_DIR}/tasks.csv"
RESULTS_CSV="${OUT_DIR}/results.csv"
RECONCILE_CSV="${OUT_DIR}/journal_reconcile.csv"
COMMANDS_TXT="${OUT_DIR}/commands.txt"
{
echo "task_key,bucket,object,version_id,storage_class,queued_at_unix_nanos,task_object"
} > "$TASKS_CSV"
{
echo "task_key,result,completion_reason,completed_at_unix_nanos,result_object"
} > "$RESULTS_CSV"
{
echo "task_key,bucket,object,version_id,storage_class,task_has_result,result,result_reason,task_job_id_match,result_job_id_match,task_path,result_path"
} > "$RECONCILE_CSV"
}
setup_mc_alias() {
if [[ "$MC_ALIAS" == "" ]]; then
MC_ALIAS="mt-journal-audit-${JOB_ID}"
MC_ALIAS_CREATED=true
if [[ "$DRY_RUN" == true ]]; then
echo "[DRY-RUN] mc alias set ${MC_ALIAS} ${MC_ENDPOINT} ${MC_ACCESS_KEY} ***"
else
"${MC_BIN}" alias set "${MC_ALIAS}" "${MC_ENDPOINT}" "${MC_ACCESS_KEY}" "${MC_SECRET_KEY}" >/dev/null
fi
fi
if ! [[ "$DRY_RUN" == true ]]; then
"${MC_BIN}" ls "${MC_ALIAS}/${SYS_BUCKET}" >/dev/null
fi
trap cleanup_mc_alias EXIT
}
api_get() {
local path="$1"
local url="${ENDPOINT%/}${path}"
if [[ -n "$ADMIN_TOKEN" ]]; then
curl -sS -H "Authorization: Bearer ${ADMIN_TOKEN}" "$url"
else
curl -sS "$url"
fi
}
build_admin_metrics_path() {
local path="/rustfs/admin/v3/metrics?types=${METRIC_TYPES}&n=${METRIC_SAMPLES}"
if [[ -n "$METRIC_INTERVAL" ]]; then
path="${path}&interval=${METRIC_INTERVAL}"
fi
printf '%s' "$path"
}
run_json_stream() {
local query="$1"
local filter="$2"
local log_lines; log_lines="$(mktemp)"
local log_errors; log_errors="$(mktemp)"
local log_paths=()
local match
while IFS= read -r match; do
[[ -n "$match" ]] && log_paths+=("$match")
done < <(compgen -G "$LOG_GLOB" || true)
if [[ "${#log_paths[@]}" -eq 0 ]]; then
log_paths=("$LOG_GLOB")
fi
if ! rg -n --color=never -- "$query" -- "${log_paths[@]}" >"$log_lines" 2>"$log_errors"; then
:
fi
if [[ -s "$log_errors" ]]; then
sed 's/^/DEBUG: /' "$log_errors"
fi
if [[ -s "$log_lines" ]]; then
jq -R "$filter" <"$log_lines"
else
echo "[]"
fi
rm -f "$log_lines" "$log_errors"
}
run_worker_logs() {
local base_filter='fromjson? | select(.event == "lifecycle_worker_state")'
local manual_filter='fromjson? | select(.event == "lifecycle_worker_state" and .job_id == env.JOB_ID and (.state | startswith("manual_transition_")))'
local dist_filter='fromjson? | select(.event == "lifecycle_worker_state" and (.state | startswith("manual_transition_")) ) | .state'
if [[ -n "$TASK_BUCKET_FILTER" ]]; then
manual_filter='fromjson? | select(.event == "lifecycle_worker_state" and .job_id == env.JOB_ID and (.state | startswith("manual_transition_")) and .bucket == env.TASK_BUCKET_FILTER)'
dist_filter='fromjson? | select(.event == "lifecycle_worker_state" and (.state | startswith("manual_transition_")) and .bucket == env.TASK_BUCKET_FILTER) | .state'
fi
echo "## lifecycle_worker_state all"
run_json_stream "lifecycle_worker_state" "$base_filter" || true
echo
echo "## lifecycle_worker_state manual_transition + job filter"
JOB_ID="$JOB_ID" TASK_BUCKET_FILTER="$TASK_BUCKET_FILTER" run_json_stream "lifecycle_worker_state" "$manual_filter" || true
echo
echo "## lifecycle_worker_state state distribution"
JOB_ID="$JOB_ID" TASK_BUCKET_FILTER="$TASK_BUCKET_FILTER" run_json_stream "lifecycle_worker_state" "$dist_filter" | jq -r . | sort | uniq -c | sort -nr || true
}
run_admin_job() {
echo "## manual transition job status"
api_get "/rustfs/admin/v3/ilm/transition/jobs/${JOB_ID}" \
| jq '{status, mode, job_id, report: .report, queue_snapshot: .queue_snapshot, cancel_requested}'
}
run_admin_events() {
echo "## admin_ilm_transition_state events"
local filter='fromjson? | select(.event == "admin_ilm_transition_state" and .job_id == env.JOB_ID) | {timestamp, operation: .operation, state: .state, result: .result, failure_reason: .failure_reason, bucket: .bucket, prefix: .prefix}'
if [[ -n "$TASK_BUCKET_FILTER" ]]; then
filter='fromjson? | select(.event == "admin_ilm_transition_state" and .job_id == env.JOB_ID and .bucket == env.TASK_BUCKET_FILTER) | {timestamp, operation: .operation, state: .state, result: .result, failure_reason: .failure_reason, bucket: .bucket, prefix: .prefix}'
fi
JOB_ID="$JOB_ID" TASK_BUCKET_FILTER="$TASK_BUCKET_FILTER" \
run_json_stream "admin_ilm_transition_state" "$filter" || true
}
run_tier_operation_failures() {
echo "## lifecycle_tier_operation_failed events"
local filter='fromjson? | select(.event == "lifecycle_tier_operation_failed") | {timestamp, bucket: .bucket, object: .object, version_id: .version_id, tier: .tier, operation: .operation, error: .error}'
if [[ -n "$TASK_BUCKET_FILTER" ]]; then
filter='fromjson? | select(.event == "lifecycle_tier_operation_failed" and .bucket == env.TASK_BUCKET_FILTER) | {timestamp, bucket: .bucket, object: .object, version_id: .version_id, tier: .tier, operation: .operation, error: .error}'
fi
JOB_ID="$JOB_ID" TASK_BUCKET_FILTER="$TASK_BUCKET_FILTER" \
run_json_stream "lifecycle_tier_operation_failed" "$filter" || true
}
run_metrics() {
echo "## metrics: scanner.lifecycle_transition"
local path
path="$(build_admin_metrics_path)"
api_get "$path" | jq '.aggregated.scanner.lifecycle_transition'
}
journal_key_prefix() {
local prefix="$1"
local shard_a="${JOB_ID:0:2}"
local shard_b="${JOB_ID:2:2}"
printf '%s/%s/%s/%s' "$prefix" "$shard_a" "$shard_b" "$JOB_ID"
}
csv_escape() {
local value="${1:-}"
value="${value//\"/\"\"}"
printf '\"%s\"' "$value"
}
read_json_file() {
local object_key="$1"
if [[ "$DRY_RUN" == true ]]; then
echo ""
return 0
fi
"${MC_BIN}" cat "${MC_ALIAS}/${SYS_BUCKET}/${object_key}"
}
collect_journals() {
if [[ "$DRY_RUN" == true ]]; then
echo "[DRY-RUN] skip .rustfs.sys object reads."
echo "dry_run=true" > "$SUMMARY_TXT"
return 0
fi
local task_prefix result_prefix
task_prefix="$(journal_key_prefix "$TASK_PREFIX")"
result_prefix="$(journal_key_prefix "$RESULT_PREFIX")"
local task_objs
local result_objs
local object_name
local task_key
local raw
local record_task_job record_result_job
local expected_job
expected_job="$JOB_ID"
local -A task_bucket task_object task_version task_storage task_queued task_job task_path
local -A result_status result_reason result_completed result_job result_path
local -A task_seen
local -A result_seen
task_objs="$("${MC_BIN}" --json ls --recursive "${MC_ALIAS}/${SYS_BUCKET}/${task_prefix}" | jq -r 'select(.type == "file") | .key')"
result_objs="$("${MC_BIN}" --json ls --recursive "${MC_ALIAS}/${SYS_BUCKET}/${result_prefix}" | jq -r 'select(.type == "file") | .key')"
while IFS= read -r object_name; do
[[ -z "$object_name" ]] && continue
task_key="${object_name##*/}"
task_key="${task_key%.json}"
if ! [[ "$task_key" =~ ^[0-9a-f]{64}$ ]]; then
echo "WARN: ignore invalid task object name: ${object_name}" >&2
continue
fi
raw="$(read_json_file "$object_name")"
if ! record_task_job="$(printf '%s' "$raw" | jq -r '.record.job_id // empty' 2>/dev/null)"; then
echo "WARN: parse error task object: ${object_name}" >&2
continue
fi
if [[ -z "$record_task_job" ]]; then
echo "WARN: task object missing record.job_id: ${object_name}" >&2
continue
fi
task_seen["$task_key"]="1"
task_bucket["$task_key"]="$(printf '%s' "$raw" | jq -r '.record.bucket // empty' )"
task_object["$task_key"]="$(printf '%s' "$raw" | jq -r '.record.object // empty' )"
task_version["$task_key"]="$(printf '%s' "$raw" | jq -r '.record.version_id // empty' )"
task_storage["$task_key"]="$(printf '%s' "$raw" | jq -r '.record.storage_class // empty' )"
task_queued["$task_key"]="$(printf '%s' "$raw" | jq -r '.record.queued_at_unix_nanos // empty' )"
task_job["$task_key"]="$(printf '%s' "$record_task_job")"
task_path["$task_key"]="$object_name"
printf '%s,%s,%s,%s,%s,%s,%s\n' \
"$(csv_escape "$task_key")" \
"$(csv_escape "${task_bucket[$task_key]}")" \
"$(csv_escape "${task_object[$task_key]}")" \
"$(csv_escape "${task_version[$task_key]}")" \
"$(csv_escape "${task_storage[$task_key]}")" \
"$(csv_escape "${task_queued[$task_key]}")" \
"$(csv_escape "$object_name")" >> "$TASKS_CSV"
done <<< "$task_objs"
while IFS= read -r object_name; do
[[ -z "$object_name" ]] && continue
task_key="${object_name##*/}"
task_key="${task_key%.json}"
if ! [[ "$task_key" =~ ^[0-9a-f]{64}$ ]]; then
echo "WARN: ignore invalid result object name: ${object_name}" >&2
continue
fi
raw="$(read_json_file "$object_name")"
if ! record_result_job="$(printf '%s' "$raw" | jq -r '.record.job_id // empty' 2>/dev/null)"; then
echo "WARN: parse error result object: ${object_name}" >&2
continue
fi
if [[ -z "$record_result_job" ]]; then
echo "WARN: result object missing record.job_id: ${object_name}" >&2
continue
fi
result_seen["$task_key"]="1"
result_status["$task_key"]="$(printf '%s' "$raw" | jq -r '.record.result // .result // empty' )"
result_reason["$task_key"]="$(printf '%s' "$raw" | jq -r '.record.failure_reason // empty' )"
result_completed["$task_key"]="$(printf '%s' "$raw" | jq -r '.record.completed_at_unix_nanos // empty' )"
result_job["$task_key"]="$(printf '%s' "$record_result_job")"
result_path["$task_key"]="$object_name"
printf '%s,%s,%s,%s,%s\n' \
"$(csv_escape "$task_key")" \
"$(csv_escape "${result_status[$task_key]}")" \
"$(csv_escape "${result_reason[$task_key]}")" \
"$(csv_escape "${result_completed[$task_key]}")" \
"$(csv_escape "$object_name")" >> "$RESULTS_CSV"
done <<< "$result_objs"
local task_count=0
local result_count=0
local missing_count=0
local mismatch_count=0
local completed_count=0
local failed_count=0
local task_job_id_match
local result_job_id_match
local task_record_job
local result_record_job
local -A failure_by_reason=()
task_count="${#task_seen[@]}"
result_count="${#result_seen[@]}"
for task_key in "${!task_seen[@]}"; do
local result=""
local reason=""
local task_has_result="false"
if [[ -n "${result_seen[$task_key]+x}" ]]; then
task_has_result="true"
result="${result_status[$task_key]}"
reason="${result_reason[$task_key]}"
if [[ "${result}" == "Completed" ]]; then
completed_count=$((completed_count + 1))
elif [[ "${result}" == "TierFailure" ]]; then
failed_count=$((failed_count + 1))
reason="${reason:-Unknown}"
failure_by_reason["$reason"]=$(( ${failure_by_reason["$reason"]:-0} + 1 ))
fi
else
missing_count=$((missing_count + 1))
fi
task_job_id_match="false"
result_job_id_match="false"
task_record_job="${task_job[$task_key]-}"
if [[ -n "$task_record_job" && "$(normalize_uuid "$task_record_job")" == "$expected_job" ]]; then
task_job_id_match="true"
fi
result_record_job="${result_job[$task_key]-}"
if [[ -n "$result_record_job" && "$(normalize_uuid "$result_record_job")" == "$expected_job" ]]; then
result_job_id_match="true"
fi
if [[ "$task_job_id_match" != "true" ]] || [[ -n "${result_seen[$task_key]+x}" && "$result_job_id_match" != "true" ]]; then
mismatch_count=$((mismatch_count + 1))
fi
printf '%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s\n' \
"$(csv_escape "$task_key")" \
"$(csv_escape "${task_bucket[$task_key]}")" \
"$(csv_escape "${task_object[$task_key]}")" \
"$(csv_escape "${task_version[$task_key]}")" \
"$(csv_escape "${task_storage[$task_key]}")" \
"$(csv_escape "$task_has_result")" \
"$(csv_escape "${result_status[$task_key]}")" \
"$(csv_escape "${result_reason[$task_key]}")" \
"$(csv_escape "$task_job_id_match")" \
"$(csv_escape "$result_job_id_match")" \
"$(csv_escape "${task_path[$task_key]}")" \
"$(csv_escape "${result_path[$task_key]-}")" >> "$RECONCILE_CSV"
done
local orphan_count=0
for task_key in "${!result_seen[@]}"; do
if [[ -z "${task_seen[$task_key]+x}" ]]; then
orphan_count=$((orphan_count + 1))
printf 'orphan_result_task_key=%s\n' "$task_key" >> "$SUMMARY_TXT"
fi
done
{
echo "job_id=$JOB_ID"
echo "sys_bucket=$SYS_BUCKET"
echo "expected_task_prefix=$task_prefix"
echo "expected_result_prefix=$result_prefix"
echo "task_count=${task_count}"
echo "result_count=${result_count}"
echo "missing_result_count=${missing_count}"
echo "job_id_mismatch_count=${mismatch_count}"
echo "orphan_result_count=${orphan_count}"
echo "completed_result_count=${completed_count}"
echo "failed_result_count=${failed_count}"
for reason in "${!failure_by_reason[@]}"; do
echo "result_failure_${reason// /_}=${failure_by_reason[$reason]}"
done
} > "$SUMMARY_TXT"
{
echo
echo "## journal summary"
echo "- tasks: ${task_count}"
echo "- results: ${result_count}"
echo "- missing result for task: ${missing_count}"
echo "- orphaned result rows: ${orphan_count}"
echo "- completed result: ${completed_count}"
echo "- failed result: ${failed_count}"
} >> "$SUMMARY_TXT"
}
build_command_notes() {
{
echo "# Run snippets (to use manually)"
echo
echo "curl -sS ${ENDPOINT}/rustfs/admin/v3/metrics?types=${METRIC_TYPES}&n=${METRIC_SAMPLES} | jq '.aggregated.scanner.lifecycle_transition'"
echo "scripts/manual_transition_journal_audit.sh --endpoint ${ENDPOINT} --job-id ${JOB_ID} --admin-token <token> --sys-bucket ${SYS_BUCKET}"
echo "scripts/manual_transition_journal_audit.sh --endpoint ${ENDPOINT} --job-id ${JOB_ID} --admin-token <token> --sys-bucket ${SYS_BUCKET} --out-dir ${OUT_DIR}"
if [[ "$DRY_RUN" == true ]]; then
echo "# Add --mc-endpoint, --mc-access-key, --mc-secret-key when writing to .rustfs.sys."
fi
} > "$COMMANDS_TXT"
}
main() {
parse_args "$@"
if [[ "$SHOW_HELP" == true ]]; then
usage
exit 0
fi
require_bash4
validate_args
require_cmd "$MC_BIN"
require_cmd curl
require_cmd jq
require_cmd rg
setup_output
setup_mc_alias
build_command_notes
run_worker_logs
echo
run_admin_events
echo
run_tier_operation_failures
echo
run_admin_job
echo
run_metrics
echo
collect_journals
echo "Artifacts: $OUT_DIR"
echo " - $TASKS_CSV"
echo " - $RESULTS_CSV"
echo " - $RECONCILE_CSV"
echo " - $SUMMARY_TXT"
echo " - $COMMANDS_TXT"
}
main "$@"
+291
View File
@@ -0,0 +1,291 @@
#!/usr/bin/env bash
# Copyright 2026 RustFS Team
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
PROJECT_ROOT="$(cd "${SCRIPT_DIR}/.." && pwd)"
ENDPOINT=""
ADMIN_TOKEN=""
OUT_DIR=""
PHASE_MATRIX="request-only:100:0:30,canary:90:10:120,mixed:50:50:240,full-rollout:0:100:360,rollback:100:0:30"
CONCURRENCIES="8,16,32"
OBJECT_COUNTS="5k,20k,50k"
JOB_BUCKET="manual-transition"
JOB_PREFIX="journal-mixed-rollout"
READ_RATIO="90"
RUN_ADMIN_CHECKS=true
DRY_RUN=false
usage() {
cat <<'USAGE'
Usage:
scripts/manual_transition_mixed_rollout_matrix.sh --endpoint <admin-api> [options]
Required:
--endpoint Admin API base, e.g. https://127.0.0.1:9000
Optional:
--admin-token Bearer token for admin calls
--phase-matrix Comma-separated phase spec: name:old_pct:new_pct:duration_min
--concurrencies Comma-separated concurrency list
--object-counts Comma-separated queue-size list (for plan rows)
--job-bucket Manual transition source bucket for sample commands
--job-prefix Prefix for generated job notes
--read-ratio Target read ratio percentage for mixed workload
--no-admin-checks Skip generated admin check commands
--out-dir Output dir
--dry-run
--help
Examples:
scripts/manual_transition_mixed_rollout_matrix.sh \
--endpoint https://127.0.0.1:9000 \
--admin-token "$TOKEN" \
--phase-matrix "request-only:100:0:30,canary:90:10:120,full:0:100:240"
USAGE
}
arg_value() {
local flag="$1"
local value="${2:-}"
if [[ -z "$value" || "$value" == --* ]]; then
echo "ERROR: missing value for $flag" >&2
exit 1
fi
printf '%s' "$value"
}
require_cmd() {
local cmd="$1"
if ! command -v "$cmd" >/dev/null 2>&1; then
echo "ERROR: command not found: $cmd" >&2
exit 1
fi
}
parse_args() {
while [[ $# -gt 0 ]]; do
case "$1" in
--endpoint) ENDPOINT="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--admin-token) ADMIN_TOKEN="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--phase-matrix) PHASE_MATRIX="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--concurrencies) CONCURRENCIES="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--object-counts) OBJECT_COUNTS="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--job-bucket) JOB_BUCKET="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--job-prefix) JOB_PREFIX="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--read-ratio) READ_RATIO="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--out-dir) OUT_DIR="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--no-admin-checks) RUN_ADMIN_CHECKS=false; shift ;;
--dry-run) DRY_RUN=true; shift ;;
-h|--help) usage; exit 0 ;;
*)
echo "ERROR: unknown arg: $1" >&2
usage
exit 1
;;
esac
done
}
trim() {
echo "$1" | awk '{$1=$1;print}'
}
parse_phase_row() {
local spec="$1"
local name old_ratio new_ratio duration_min
name="$(echo "$spec" | cut -d':' -f1)"
old_ratio="$(echo "$spec" | cut -d':' -f2)"
new_ratio="$(echo "$spec" | cut -d':' -f3)"
duration_min="$(echo "$spec" | cut -d':' -f4)"
name="$(trim "$name")"
old_ratio="$(trim "$old_ratio")"
new_ratio="$(trim "$new_ratio")"
duration_min="$(trim "$duration_min")"
if [[ -z "$name" || -z "$old_ratio" || -z "$new_ratio" || -z "$duration_min" ]]; then
echo "ERROR: invalid phase spec: $spec" >&2
exit 1
fi
if ! [[ "$old_ratio" =~ ^[0-9]+$ && "$new_ratio" =~ ^[0-9]+$ && "$duration_min" =~ ^[0-9]+$ ]]; then
echo "ERROR: phase ratio/duration must be integers: $spec" >&2
exit 1
fi
echo "$name|$old_ratio|$new_ratio|$duration_min"
}
gate_for_phase() {
case "$1" in
request-only)
echo "No mixed-version errors; unknown failures should stay 0 in job report; log ratio must trend down."
;;
canary)
echo "old/new mismatch in worker results = 0; no Unknown job failures."
;;
mixed)
echo "request-only + canary acceptance gates both pass for 2+ windows, then proceed."
;;
full-rollout|full)
echo "full canary-to-full transition completed, no tier_failure in report, worker_result mismatch = 0."
;;
rollback)
echo "all rollout checks reverse cleanly; queued -> completed in job terminal state."
;;
*)
echo "run admin job status + journal reconcile + metric delta sanity check."
;;
esac
}
admin_check_cmd() {
local phase="$1"
local job_id_ref="$2"
local metric_types="$3"
local metric_samples="$4"
local metric_path
local job_url
local metric_url
local job_filter='.report, .queue_snapshot, .failure_reason'
local metric_filter='.aggregated.scanner.lifecycle_transition'
if [[ "$RUN_ADMIN_CHECKS" != true ]]; then
return
fi
metric_path="/rustfs/admin/v3/metrics?types=${metric_types}&n=${metric_samples}"
job_url="${ENDPOINT%/}/rustfs/admin/v3/ilm/transition/jobs/${job_id_ref}"
metric_url="${ENDPOINT%/}${metric_path}"
{
printf '# check-%s\n' "$phase"
printf 'curl -sS'
if [[ -n "$ADMIN_TOKEN" ]]; then
printf " -H \"Authorization: Bearer \${ADMIN_TOKEN}\""
fi
printf ' -X GET "%s" | jq '\''%s'\''\n' "$job_url" "$job_filter"
printf 'curl -sS'
if [[ -n "$ADMIN_TOKEN" ]]; then
printf " -H \"Authorization: Bearer \${ADMIN_TOKEN}\""
fi
printf ' -X GET "%s" | jq '\''%s'\''\n' "$metric_url" "$metric_filter"
printf './scripts/manual_transition_journal_audit.sh --endpoint %s --job-id "%s"' "$ENDPOINT" "$job_id_ref"
if [[ -n "$ADMIN_TOKEN" ]]; then
printf " --admin-token \"\${ADMIN_TOKEN}\""
fi
printf ' --sys-bucket .rustfs.sys\n'
}
}
run_rows() {
local matrix_csv="${OUT_DIR}/mixed_rollout_matrix.csv"
local run_script="${OUT_DIR}/run_phase_commands.sh"
local checklist="${OUT_DIR}/mixed_rollout_checklist.md"
local command_manifest="${OUT_DIR}/run_phase_notes.txt"
local admin_check
local admin_check_csv
local job_id_ref
{
echo "phase,old_pct,new_pct,duration_min,concurrency,object_count,read_ratio,gate,admin_check"
} > "$matrix_csv"
{
echo "#!/usr/bin/env bash"
echo "set -euo pipefail"
echo ""
echo "# generated matrix commands: fill in JOB_ID after each manual transition run"
echo ""
} > "$run_script"
{
echo "# mixed-version rollout command template"
echo "Endpoint: ${ENDPOINT}"
echo "Bucket: ${JOB_BUCKET}"
echo "Prefix: ${JOB_PREFIX}"
echo "Read ratio: ${READ_RATIO}%"
echo ""
} > "$checklist"
while IFS=',' read -r phase_spec; do
[[ -z "$phase_spec" ]] && continue
parsed="$(parse_phase_row "$phase_spec")"
IFS='|' read -r phase_name old_ratio new_ratio duration_min <<< "$parsed"
gate="$(gate_for_phase "$phase_name")"
IFS=',' read -r -a counts <<< "$OBJECT_COUNTS"
IFS=',' read -r -a concs <<< "$CONCURRENCIES"
for conc in "${concs[@]}"; do
conc="$(trim "$conc")"
[[ -z "$conc" ]] && continue
for count in "${counts[@]}"; do
count="$(trim "$count")"
[[ -z "$count" || -z "$conc" ]] && continue
job_id_ref="<JOB_ID_${phase_name}_c${conc}_q${count}>"
admin_check="$(admin_check_cmd "$phase_name" "$job_id_ref" 1 1 | tr '\n' ';')"
admin_check_csv="${admin_check//\"/\"\"}"
echo "${phase_name},${old_ratio},${new_ratio},${duration_min},${conc},${count},${READ_RATIO},\"${gate}\",\"${admin_check_csv}\"" >> "$matrix_csv"
{
echo "# phase: ${phase_name} concurrency: ${conc} object_count: ${count}"
echo "# old:new = ${old_ratio}:${new_ratio}, duration=${duration_min}min"
printf './scripts/manual_transition_journal_audit.sh --endpoint %s --job-id "%s"' "$ENDPOINT" "$job_id_ref"
if [[ -n "$ADMIN_TOKEN" ]]; then
printf " --admin-token \"\${ADMIN_TOKEN}\""
fi
printf '\n'
echo ""
} >> "$run_script"
done
done
done < <(tr ',' '\n' <<< "$PHASE_MATRIX")
{
echo "## Mixed-version rollout matrix artifacts"
echo ""
echo "- matrix: ${matrix_csv}"
echo "- commands: ${run_script}"
echo "- checklist: ${checklist}"
} > "$command_manifest"
}
main() {
parse_args "$@"
if [[ -z "$ENDPOINT" ]]; then
echo "ERROR: --endpoint is required" >&2
usage
exit 1
fi
if [[ -z "$OUT_DIR" ]]; then
OUT_DIR="${PROJECT_ROOT}/target/manual-transition-mixed-rollout/$(date +%Y%m%dT%H%M%S)"
fi
mkdir -p "$OUT_DIR"
require_cmd awk
run_rows
if [[ "$DRY_RUN" == true ]]; then
echo "[DRY-RUN] generated files:"
fi
echo "Generated:"
echo " - ${OUT_DIR}/mixed_rollout_matrix.csv"
echo " - ${OUT_DIR}/run_phase_commands.sh"
echo " - ${OUT_DIR}/mixed_rollout_checklist.md"
echo " - ${OUT_DIR}/run_phase_notes.txt"
}
main "$@"
+262
View File
@@ -0,0 +1,262 @@
#!/usr/bin/env bash
# Copyright 2026 RustFS Team
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
# http://www.apache.org/licenses/LICENSE-2.0
#
# software distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
PROJECT_ROOT="$(cd "${SCRIPT_DIR}/.." && pwd)"
ENDPOINT=""
ADMIN_TOKEN=""
OUT_DIR=""
WINDOW_SPEC="nightly-2h:120:16:5000:balanced,nightly-12h:720:24:8000:write-heavy,nightly-24h:1440:32:12000:read-heavy"
SOAK_RATIOS="read-heavy:90:10,balanced:70:30,write-heavy:40:60"
WORKLOAD_SIZES="4KiB,1MiB,16MiB"
RUN_ADMIN_CHECKS=true
COST_BUDGET="$((24*1024))"
DRY_RUN=false
usage() {
cat <<'USAGE'
Usage:
scripts/manual_transition_soak_matrix.sh --endpoint <admin-api> [options]
Required:
--endpoint Admin API base
Optional:
--admin-token Bearer token for admin calls
--window-spec CSV of <name:duration_min:concurrency:ops_per_minute:mix_name>
--workload-sizes Comma-separated object size set
--soak-ratios Comma-separated mix spec: label:read_pct:write_pct
--cost-budget Max budget units for each run, default 24576
--out-dir Output dir
--no-admin-checks Skip admin check command blocks
--dry-run
--help
USAGE
}
arg_value() {
local flag="$1"
local value="${2:-}"
if [[ -z "$value" || "$value" == --* ]]; then
echo "ERROR: missing value for $flag" >&2
exit 1
fi
printf '%s' "$value"
}
require_cmd() {
local cmd="$1"
if ! command -v "$cmd" >/dev/null 2>&1; then
echo "ERROR: command not found: $cmd" >&2
exit 1
fi
}
parse_args() {
while [[ $# -gt 0 ]]; do
case "$1" in
--endpoint) ENDPOINT="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--admin-token) ADMIN_TOKEN="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--window-spec) WINDOW_SPEC="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--workload-sizes) WORKLOAD_SIZES="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--soak-ratios) SOAK_RATIOS="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--cost-budget) COST_BUDGET="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--out-dir) OUT_DIR="$(arg_value "$1" "${2:-}")"; shift 2 ;;
--no-admin-checks) RUN_ADMIN_CHECKS=false; shift ;;
--dry-run) DRY_RUN=true; shift ;;
-h|--help) usage; exit 0 ;;
*)
echo "ERROR: unknown arg: $1" >&2
usage
exit 1
;;
esac
done
}
trim() {
echo "$1" | awk '{$1=$1;print}'
}
parse_window() {
local spec="$1"
local name duration_min concurrency ops_per_min mix_name
name="$(echo "$spec" | cut -d':' -f1)"
duration_min="$(echo "$spec" | cut -d':' -f2)"
concurrency="$(echo "$spec" | cut -d':' -f3)"
ops_per_min="$(echo "$spec" | cut -d':' -f4)"
mix_name="$(echo "$spec" | cut -d':' -f5)"
name="$(trim "$name")"
duration_min="$(trim "$duration_min")"
concurrency="$(trim "$concurrency")"
ops_per_min="$(trim "$ops_per_min")"
mix_name="$(trim "$mix_name")"
if [[ -z "$name" || -z "$duration_min" || -z "$concurrency" || -z "$ops_per_min" || -z "$mix_name" ]]; then
echo "ERROR: invalid window spec: $spec" >&2
exit 1
fi
if ! [[ "$duration_min" =~ ^[0-9]+$ && "$concurrency" =~ ^[0-9]+$ && "$ops_per_min" =~ ^[0-9]+$ ]]; then
echo "ERROR: window duration/concurrency/ops must be integers: $spec" >&2
exit 1
fi
echo "$name|$duration_min|$concurrency|$ops_per_min|$mix_name"
}
parse_ratio() {
local spec="$1"
local name read_pct write_pct
name="$(echo "$spec" | cut -d':' -f1)"
read_pct="$(echo "$spec" | cut -d':' -f2)"
write_pct="$(echo "$spec" | cut -d':' -f3)"
name="$(trim "$name")"
read_pct="$(trim "$read_pct")"
write_pct="$(trim "$write_pct")"
if [[ -z "$name" || -z "$read_pct" || -z "$write_pct" ]]; then
echo "ERROR: invalid mix ratio: $spec" >&2
exit 1
fi
if ! [[ "$read_pct" =~ ^[0-9]+$ && "$write_pct" =~ ^[0-9]+$ ]]; then
echo "ERROR: mix percentages must be integers: $spec" >&2
exit 1
fi
echo "$name|$read_pct|$write_pct"
}
resolve_mix() {
local name="$1"
local line
line="$(awk -v target="$name" -F: '{
if ($1 == target) { print $0; exit }
}' < <(tr ',' '\n' <<< "$SOAK_RATIOS"))"
if [[ -z "$line" ]]; then
return 1
fi
parse_ratio "$line"
}
run_checks() {
local tag="$1"
local job_id="$2"
local token_hdr=""
if [[ -z "$job_id" ]]; then
job_id="<JOB_ID_${tag}>"
fi
if [[ -n "$ADMIN_TOKEN" ]]; then
token_hdr="-H \"Authorization: Bearer \${ADMIN_TOKEN}\" "
fi
if [[ "$RUN_ADMIN_CHECKS" != true ]]; then
return
fi
echo "# admin checks for ${tag}"
echo "curl -sS ${token_hdr}-X GET \"${ENDPOINT%/}/rustfs/admin/v3/ilm/transition/jobs/${job_id}\" | jq '.status, .report, .queue_snapshot, .failure_reason'"
echo "curl -sS ${token_hdr}-X GET \"${ENDPOINT%/}/rustfs/admin/v3/metrics?types=1&n=1\" | jq '.aggregated.scanner.lifecycle_transition'"
printf './scripts/manual_transition_journal_audit.sh --endpoint %s --job-id "%s"' "$ENDPOINT" "$job_id"
if [[ -n "$ADMIN_TOKEN" ]]; then
printf " --admin-token \"\${ADMIN_TOKEN}\""
fi
printf '\n'
}
write_matrix_files() {
local matrix_csv="${OUT_DIR}/nightly_soak_matrix.csv"
local run_matrix="${OUT_DIR}/run_soak_matrix.sh"
local notes="${OUT_DIR}/soak_notes.md"
local run_id=1
{
echo "window,window_duration_min,concurrency,ops_per_min,mix_name,read_pct,write_pct,size,expected_ops,expected_run_id,run_check"
} > "$matrix_csv"
echo "#!/usr/bin/env bash" > "$run_matrix"
echo "set -euo pipefail" >> "$run_matrix"
echo "" >> "$run_matrix"
{
echo "# Nightly/night soak matrix for manual transition stress"
echo "- endpoint: ${ENDPOINT}"
echo "- cost-budget: ${COST_BUDGET}"
echo ""
echo "Recommended gate for each row: no unknown failure_reason spikes, job result drift is 0, and queued/task/result counts reconcile."
echo ""
} > "$notes"
IFS=',' read -r -a sizes <<< "$WORKLOAD_SIZES"
while IFS=',' read -r window_spec; do
[[ -z "$window_spec" ]] && continue
parsed="$(parse_window "$window_spec")"
IFS='|' read -r window_name duration_min concurrency ops_per_min mix_name <<< "$parsed"
mix="$(resolve_mix "$mix_name" || true)"
if [[ -z "$mix" ]]; then
echo "ERROR: unknown mix name in window spec: ${window_name} -> ${mix_name}" >&2
exit 1
fi
IFS='|' read -r resolved_mix read_pct write_pct <<< "$mix"
for size in "${sizes[@]}"; do
size="$(trim "$size")"
[[ -z "$size" ]] && continue
expected_ops=$(( duration_min * ops_per_min ))
if (( expected_ops > COST_BUDGET )); then
budget_status="over-budget"
else
budget_status="within-budget"
fi
echo "${window_name},${duration_min},${concurrency},${ops_per_min},${resolved_mix},${read_pct},${write_pct},${size},${expected_ops},${run_id},${budget_status}" >> "$matrix_csv"
{
echo "# window=${window_name} size=${size} mix=${resolved_mix} concurrency=${concurrency} duration=${duration_min}m ops_per_min=${ops_per_min}"
echo "job_id=\"<JOB_ID_${window_name}_${size}_${concurrency}_${run_id}>\""
echo "read_pct=${read_pct}"
echo "write_pct=${write_pct}"
echo "expected_ops=${expected_ops}"
echo "budget_status=${budget_status}"
run_checks "${window_name}_${size}_${concurrency}_${run_id}" "<JOB_ID_${window_name}_${size}_${concurrency}_${run_id}>"
echo
} >> "$run_matrix"
run_id=$((run_id + 1))
done
done < <(tr ',' '\n' <<< "$WINDOW_SPEC")
{
echo "# run target:"
echo "bash ${run_matrix}"
} >> "$notes"
}
main() {
parse_args "$@"
if [[ -z "$ENDPOINT" ]]; then
echo "ERROR: --endpoint is required" >&2
usage
exit 1
fi
if [[ -z "$OUT_DIR" ]]; then
OUT_DIR="${PROJECT_ROOT}/target/manual-transition-nightly-soak/$(date +%Y%m%dT%H%M%S)"
fi
mkdir -p "$OUT_DIR"
require_cmd awk
write_matrix_files
if [[ "$DRY_RUN" == true ]]; then
echo "[DRY-RUN] generated matrix files:"
fi
echo "Generated:"
echo " - ${OUT_DIR}/nightly_soak_matrix.csv"
echo " - ${OUT_DIR}/run_soak_matrix.sh"
echo " - ${OUT_DIR}/soak_notes.md"
}
main "$@"