f52a389652
fix(ecstore): persist unresolved decommission entries ( #6415 )
...
* fix(ecstore): persist unresolved decommission entries
* fix(ecstore): type decommission completion result
* fix(ecstore): allow intentional decommission listing signatures under strict clippy
The sftp/swift feature-matrix clippy gates run with -D warnings and
flag the unresolved-entry resolver (large Err payload by design, 8
context parameters) and the decommission listing driver (9 args).
Document why and align with the existing decommission_entry precedent.
2026-08-24 14:05:08 +08:00
73cd1b5be2
fix(ecstore): fence rebalance and decommission activation ( #6400 )
...
* fix(ecstore): fence rebalance and decommission activation
* fix(ecstore): fence lost activation locks
* fix(ecstore): bind rebalance workers to activation id
* fix(ecstore): close rebalance activation races
* test(ecstore): exercise lost rebalance commit fence
* fix(ecstore): repair rebalance fence test wiring
* test(ecstore): reuse rebalance metadata fixture
* fix(ecstore): satisfy rebalance activation clippy checks
* fix(ecstore): fence stale rebalance workers
* fix(ecstore): commit rebalance activation after persistence
* fix(ecstore): fence rebalance commits and unblock stop
* test(ecstore): exercise real rebalance fences
* fix(rebalance): cancel admin stop before activation wait
* fix(ecstore): fence multipart staging on rebalance lock loss
* fix(ecstore): adopt activations after durable commit
* fix(rebalance): preserve committed activation recovery
* fix(rebalance): make prepared stop terminal-safe
* fix(ecstore): repair rebalance test imports
* fix(ecstore): repair rebalance entry runtime failures
* test(ecstore): fix activation fence synchronization
* test(ecstore): scope rebalance disk trait import
* test(ecstore): observe decommission lock attempt
* fix(ecstore): align activation fence test imports
* fix(ecstore): remove duplicate activation test import
* fix(ecstore): resolve CI clippy failures
* fix: satisfy activation merge lint gates
---------
Co-authored-by: houseme <housemecn@gmail.com >
2026-08-24 14:03:21 +08:00
40bf9f0425
fix(ecstore): prefer active pool reads during decommission ( #6475 )
2026-08-24 09:36:35 +08:00
eb0384c225
test(app): distinguish usage overlay from quota floor ( #6479 )
...
test(app): preserve delete quota floor assertion
2026-08-24 09:33:27 +08:00
0e015360cc
test(scanner): repair post-fence fixtures ( #6478 )
2026-08-24 09:33:21 +08:00
2bd1df3075
fix(scanner): preserve default usage cache wire format ( #6477 )
2026-08-24 09:33:16 +08:00
9935911e93
fix(ecstore): drop unused decommission bucket runner and refresh e2e linux digest ( #6476 )
...
fix(rustfs): drop sftp-dead scanner capability import and unwired heal_bucket trait method
The sftp feature matrix compiles fewer call sites: the plain
sign_ns_scanner_capability re-export had no remaining user, and
StoragePeerS3ClientExt::heal_bucket was superseded by
heal_bucket_with_fence when #6416 wired the fenced path.
2026-08-24 09:33:10 +08:00
99fb77b164
fix(ci): restore main checks ( #6469 )
...
* fix(ci): restore main checks
* fix(scanner): bootstrap pristine usage state
* fix(scanner): reject empty usage snapshots
2026-08-24 09:33:00 +08:00
ebff02304d
fix(connect): make registration bootstrap retry durable ( #6468 )
...
* fix(connect): make registration state durability retry-safe
* fix(connect): reject parent state paths
* fix(connect): harden state directory creation
* fix(connect): bound bootstrap directory syncs
* fix(connect): require durable state parent
* fix(connect): close bootstrap marker race
* test(connect): cover marker sync failure
2026-08-24 09:32:54 +08:00
57eaa8228d
fix(ecstore): migrate tier free versions during decommission ( #6393 )
2026-08-24 09:21:53 +08:00
aa0a374aa4
style: rustfmt decommission import groups on main ( #6474 )
...
cargo fmt --check has been failing since #6410 landed: the merged
decommission import groups in core/pools.rs were not canonical
rustfmt output. Apply formatting verbatim, no code changes.
2026-08-24 03:07:49 +08:00
1ec1a8d90d
fix(ecstore): run metadata decommission before buckets ( #6410 )
...
* fix(ecstore): run decommission metadata first
* fix(ecstore): clear clippy warnings
---------
Co-authored-by: houseme <housemecn@gmail.com >
2026-08-24 00:22:46 +08:00
28b3ecf547
fix(ci): restore main checks ( #6467 )
2026-08-24 00:15:58 +08:00
33341d5fcf
fix(ecstore): unify decommission resume queue ( #6403 )
...
* fix(ecstore): unify decommission resume queue
* fix(ecstore): clear swift clippy warnings
* fix(ecstore): serialize decommission recovery
---------
Co-authored-by: houseme <housemecn@gmail.com >
2026-08-23 23:51:33 +08:00
2a43e021c9
fix(ecstore): fence bucket heal during decommission ( #6416 )
...
* fix(ecstore): fence bucket heal during decommission
* fix(ecstore): preserve unfenced heal compatibility
---------
Co-authored-by: houseme <housemecn@gmail.com >
2026-08-23 23:38:16 +08:00
8bf4f20890
test(e2e): activate conditional write regressions ( #6454 )
2026-08-23 23:34:50 +08:00
d269b90201
fix(ci): restore main branch checks ( #6465 )
...
* fix(protocols): add missing test dependency
* style(ecstore): fix decommission formatting
2026-08-23 23:34:46 +08:00
83aa9c221b
test: add coalescer delay cost report ( #6464 )
...
Add a read-only Prometheus report helper for backlog#2007 so the 200us vs 50us coalescer delay experiment can capture RPC, batch distribution, stage latency, and host-cost signals with one fixed evidence format.
Co-authored-by: heihutu <heihutu@gmail.com >
2026-08-23 23:22:38 +08:00
7f32c675ac
fix(ecstore): recover pool metadata from replicas ( #6457 )
2026-08-23 23:22:03 +08:00
8ccf7151f3
fix(ecstore): persist cancel before signaling ( #6423 )
...
* fix(ecstore): persist decommission cancel before signaling
* test(ecstore): align cancel regressions with movement gate
---------
Co-authored-by: houseme <housemecn@gmail.com >
2026-08-23 23:17:14 +08:00
8c3aeaebed
fix(ecstore): drain multipart uploads before decommission ( #6414 )
...
* fix(ecstore): drain multipart uploads before decommission
* fix(ecstore): prioritize active multipart pools
---------
Co-authored-by: houseme <housemecn@gmail.com >
2026-08-23 23:04:35 +08:00
201c653dcd
fix(ci): restore workspace lint compatibility ( #6460 )
2026-08-23 22:35:43 +08:00
3ce01dcc73
fix(object-lock): unblock authorized replication writes on locked versions and tolerate cleared lock metadata ( #6413 )
2026-08-23 22:35:23 +08:00
0d15ce1865
ci: install protocol socket oracle ( #6411 )
...
Co-authored-by: houseme <housemecn@gmail.com >
2026-08-23 21:33:00 +08:00
415427f99d
feat(connect): emit low-frequency inventory ( #6418 )
...
* feat(connect): emit low-frequency inventory
* fix(connect): retry incomplete inventory samples
* fix(connect): reset inventory sampling backoff
* fix(connect): mark missing drives offline in inventory
* fix(connect): harden inventory snapshots
* fix(connect): validate persisted inventory state
* fix(connect): close inventory lifecycle gaps
* fix(connect): validate inventory topology slots
* fix(connect): validate inventory geometry
2026-08-23 21:32:54 +08:00
18cde00fad
fix(ecstore): retry decommission entries safely on source changes ( #6419 )
...
A single object's SourceChanged during decommission cleanup no longer
cancels the shared worker token and fails the whole pool operation.
Cleanup preflight and source-cleanup outcomes now retry per entry with
bounded attempts and cancellation-aware backoff, applied uniformly to
ordinary versions, delete markers, and tiered copies (removing the
try-once-only branches); every retry re-lists the entry and redoes
version multiset validation before touching the source. Only quorum
loss, unrecoverable system errors, or exceeding a pool-level
SourceChanged exhaustion threshold still fails the decommission, and
exhausted entries never delete their source versions.
Retry attempts, backoff, and deferred/exhausted reasons are logged per
entry for observability. Heavy regression tests spawn on dedicated
32MiB stacks following the existing store-test pattern.
Fixes rustfs/backlog#1913
Co-authored-by: houseme <housemecn@gmail.com >
2026-08-23 21:32:49 +08:00
2788ef7229
feat(connect): collect bounded offline diagnostics ( #6450 )
...
* feat(connect): collect bounded offline diagnostics
* fix(connect): tighten offline diagnostic boundaries
2026-08-23 21:32:43 +08:00
5cedf73d7c
fix(ecstore): terminate walk directory streams ( #6462 )
...
* fix(ecstore): terminate walk directory streams
* test(e2e): refresh node service selection
* fix(filemeta): initialize empty metacache streams
2026-08-23 21:32:39 +08:00
f694a0000a
fix(server): adapt quick-xml name handling ( #6458 )
2026-08-23 20:37:42 +08:00
ba8f2e90be
feat(connect): add registration bootstrap command ( #6452 )
2026-08-23 20:24:23 +08:00
43a10e3c24
fix(ecstore): tolerate migration identity rewrites ( #6447 )
2026-08-23 20:24:04 +08:00
11a90ce843
test(e2e): add platform-safe selection updates ( #6424 )
2026-08-23 20:22:46 +08:00
b52cdd69ae
test(e2e): fail incomplete conditional PUT races ( #6421 )
2026-08-23 20:20:16 +08:00
06ef472def
fix(ci): make Warp ABBA evidence bounded and complete ( #6417 )
2026-08-23 20:20:01 +08:00
d3c0714b3a
ci: add s3tests upstream HEAD canary ( #6409 )
2026-08-23 20:19:13 +08:00
b5b060a9a1
ci: pin s3tests Python tools ( #6407 )
2026-08-23 20:18:56 +08:00
5bb9ffffcd
test(e2e): enforce external client prerequisites ( #6402 )
2026-08-23 20:18:41 +08:00
6f6dd19cc7
ci(coverage): add security ratchet calibration ( #6388 )
2026-08-23 20:18:04 +08:00
31933c32f9
fix(replication): apply receiver-side LWW to inbound metadata categories ( #6379 )
2026-08-23 20:17:50 +08:00
0e88a27d05
fix(admin): use madmin key names in list-remote-targets response ( #6377 )
2026-08-23 20:17:34 +08:00
35e264a9f5
fix(admin): advertise IAM admin capabilities in runtime capabilities ( #6336 )
2026-08-23 20:17:24 +08:00
38d37121ca
test(e2e): activate group management regressions ( #6405 )
2026-08-23 20:17:07 +08:00
3f3b9fd426
perf(ecstore): attribute batch read version wait stages ( #6456 )
2026-08-23 19:31:14 +08:00
a8e4b67d99
feat(metrics): expose deferred usage freshness ( #6449 )
2026-08-23 19:31:01 +08:00
b2e60be647
fix(scanner): fence unknown tier accounting ( #6396 )
2026-08-23 19:28:43 +08:00
14e3eb787d
fix(heal): correct progress accounting ( #6382 )
...
* fix(heal): correct progress accounting
* fix(heal): atomically persist page progress
* fix(heal): preserve terminal progress counters
* fix(heal): make resume handoff crash safe
* fix(heal): preserve resumable bucket checkpoints
* fix(heal): satisfy checkpoint outcome lint
* fix(heal): preserve progress status across nodes
* fix(heal): stabilize progress generations
* style: restore rebalance formatting
* test(heal): cover cross-set baseline generation
---------
Signed-off-by: houseme <housemecn@gmail.com >
Co-authored-by: overtrue <anzhengchao@gmail.com >
Co-authored-by: houseme <housemecn@gmail.com >
Co-authored-by: heihutu <heihutu@gmail.com >
2026-08-23 17:29:45 +08:00
76a863b3ea
fix(scanner): unify unknown metadata size accounting ( #6394 )
...
* fix(scanner): unify unknown metadata size accounting
* fix(scanner): preserve restore expiry semantics
* fix(ci): resolve ecstore clippy warnings
* fix(scanner): close lifecycle review gaps
---------
Signed-off-by: houseme <housemecn@gmail.com >
Co-authored-by: houseme <housemecn@gmail.com >
2026-08-23 17:28:52 +08:00
e196a134cc
fix(scanner): fence system metadata publication ( #6444 )
...
* feat(scanner): fence usage publication during data movement
* fix(scanner): detect movement refresh state changes
* fix(scanner): fence publication during data movement
* fix(scanner): close movement epoch publication races
* fix(scanner): fence movement-sensitive publication paths
* fix(scanner): fence cache and heal recovery paths
* fix(scanner): carry publication epoch through scan cycle
* fix(scanner): recheck remote cache epoch after save
* fix(scanner): recheck local cache epoch before publish
* fix(scanner): fence data usage writers and baseline
* fix(scanner): expose decommission activity to publication fence
* fix(scanner): release publication gate before reads
* fix(scanner): complete publication fence integration
* fix(scanner): avoid empty usage baseline publication
* chore(scanner): gate test-only helpers
* fix: use decommission canceler in reload test
---------
Co-authored-by: houseme <housemecn@gmail.com >
2026-08-23 17:28:21 +08:00
9cda615519
fix(scanner): discover sub-quorum heal candidates ( #6384 )
...
* fix(scanner): preserve unversioned heal retries
* fix(scanner): bound orphan heal discovery fallback
* fix(filemeta): fence unsafe heal key components
* fix(scanner): preserve exact overflow heal versions
---------
Co-authored-by: houseme <housemecn@gmail.com >
2026-08-23 16:46:36 +08:00
9a8ca3a7a9
fix(heal): add bounded resume artifact inspection ( #6420 )
...
Co-authored-by: houseme <housemecn@gmail.com >
2026-08-23 16:45:34 +08:00
Zhengchao An