Commit Graph

91 Commits

Author SHA1 Message Date
houseme b16120dbcc feat: optimize small GET read paths (#4022) 2026-07-01 15:40:00 +08:00
cxymds c67eea7bfa fix(ecstore): preserve rename data rollback metadata (#4107) 2026-06-30 15:45:27 +08:00
Zhengchao An a7dd9603e9 refactor(ecstore): add mmap copy disk read name (#4061) 2026-06-29 21:18:41 +08:00
Zhengchao An 710ae74cde perf: add S3 operations benchmark framework (#738) (#4005) 2026-06-28 18:02:41 +08:00
Zhengchao An c475d03b48 fix: replace unwrap() with expect() in more files (#729 batch 14) (#3994) 2026-06-28 11:45:23 +08:00
houseme 27468ebfa9 feat(get): consolidate GET performance optimization (#3972)
* feat(get): consolidate GET performance optimization

Consolidated implementation of all GET performance optimizations into
a single, well-organized commit replacing the previous patch-on-patch
approach.

## Changes

### Configuration (set_disk/mod.rs)
- Consolidated all GET optimization flags into a single organized section
- Enabled by default: codec streaming, metadata early-stop, page cache reclaim
- Added codec streaming multipart flag (default: disabled)
- Added version-aware early-stop flag (default: disabled)
- Added adaptive duplex buffer sizing based on object size
- All flags use OnceLock caching with rollout percentage support

### Metadata Early-Stop (set_disk/read.rs)
- Delete marker early-stop when quorum agrees
- Version-aware early-stop for versioned GET requests
- MetadataQuorumAccumulator enhanced with:
  - delete_marker_votes tracking
  - requested_version_id and matching_version_votes tracking
  - version_early_stop_decision() method
- 6 new tests for version early-stop scenarios

### Codec Streaming (erasure/coding/decode_reader.rs)
- DualInFlight (2-stripe lookahead) enabled by default

### Decode Pipeline (erasure/coding/decode.rs)
- Stripe prefetch count configuration
- Bitrot-decode overlap configuration

### Disk Layer (disk/local.rs)
- O_DIRECT read configuration constants (preparation)

### Metrics (io-metrics/lib.rs)
- BytesPool acquisition/return metrics
- Metadata phase duration with early-stop label
- Total duration with reader_path label

### Diagnostics (diagnostics/)
- Early-stop reason constants
- Pool tier/outcome label constants

### Observability (.docker/observability/)
- 3 Grafana dashboards for GET optimization monitoring
- Prometheus alert rules (6 alerts: 3 critical, 3 warning)
- Updated README.md and README_ZH.md with usage docs

### Config (config/src/constants/runtime.rs)
- Page cache reclaim read enabled by default

## Environment Variables

| Variable | Default | Description |
|----------|---------|-------------|
| RUSTFS_GET_CODEC_STREAMING_ENABLE | true | Codec streaming base flag |
| RUSTFS_GET_CODEC_STREAMING_ROLLOUT_PCT | 100 | Codec streaming rollout % |
| RUSTFS_GET_CODEC_STREAMING_MULTIPART_ENABLE | false | Multipart codec streaming |
| RUSTFS_GET_METADATA_EARLY_STOP_ENABLE | true | Early-stop base flag |
| RUSTFS_GET_METADATA_EARLY_STOP_ROLLOUT_PCT | 100 | Early-stop rollout % |
| RUSTFS_GET_METADATA_VERSION_EARLY_STOP_ENABLE | false | Version-aware early-stop |
| RUSTFS_OBJECT_FILE_CACHE_RECLAIM_READ_ENABLE | true | Page cache reclaim |
| RUSTFS_OBJECT_DIRECT_IO_READ_ENABLE | false | O_DIRECT (preparation) |
| RUSTFS_GET_DECODE_STRIPE_PREFETCH_COUNT | 1 | Stripe prefetch |
| RUSTFS_GET_BITROT_DECODE_OVERLAP_ENABLE | false | Bitrot-decode overlap |
| RUSTFS_GET_CODEC_STREAMING_MAX_INFLIGHT | 2 | DualInFlight stripes |

## Rollback

All optimizations can be disabled via environment variables:
RUSTFS_GET_CODEC_STREAMING_ENABLE=false
RUSTFS_GET_METADATA_EARLY_STOP_ENABLE=false
RUSTFS_OBJECT_FILE_CACHE_RECLAIM_READ_ENABLE=false

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

* test(get): add stress test scripts for GET optimization validation

- quick-validate-get-optimization.sh: Quick 5-minute validation
- stress-test-get-optimization.sh: Full 30+ minute stress test
- README-stress-test.md: Usage documentation

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

* test(ecstore): align file cache reclaim defaults

* chore(deps): update redis and erasure codec

* test(ecstore): align decode fill policy default

* test(ecstore): align metadata early-stop default

* fix(ecstore): keep metadata early stop opt-in

---------

Co-authored-by: heihutu <heihutu@gmail.com>
2026-06-28 07:14:07 +08:00
GatewayJ 675597ec16 fix(ecstore): handle stalled recovery reads and listings (#3790)
* fix(ecstore): handle stalled recovery reads and listings

* fix(rio): start HTTP stall timeout on read

* fix(ecstore): handle stalled reads and partial lists

* fix(ecstore): retire stalled shards and list errors

* fix(ecstore): preserve list merge lookahead entries

* fix(ecstore): bound zero-copy shard reads

* fix(ecstore): hedge stalled shard reads

* fix(ecstore): retire abandoned shard reads

* fix(ecstore): include part identity in metadata quorum

* fix(ecstore): validate heal shard sources

* fix(ecstore): verify reconstructed read shards

* chore(ecstore): log slow object read stages

* fix(heal): throttle auto heal during recovery

* fix(scanner): yield to foreground reads

* fix(scanner): track streaming object reads

* fix(ecstore): avoid false read heal fanout

* fix(ecstore): verify codec streaming reconstruction sources

* fix(ecstore): preserve quorum progress on slow shards

* fix(storage): restore read timeout facade

* fix(ecstore): retain fallback readers after quorum

* chore: allow decode helper argument lists

---------

Co-authored-by: overtrue <anzhengchao@gmail.com>
2026-06-27 10:21:09 +08:00
Zhengchao An 0a5b1b1b3a refactor: consolidate ecstore owner module layout (#3934)
* refactor: shrink ecstore root owner facades

* refactor: remove ecstore core store root shims

* refactor: move ecstore erasure owner modules

* refactor: remove ecstore root rpc facade

* refactor: move ecstore services domain modules
2026-06-27 09:03:20 +08:00
Zhengchao An 7e60432588 refactor: centralize ecstore runtime owner sources (#3798) 2026-06-24 06:34:06 +08:00
houseme a53ccb2407 fix(ecstore): reject symlink path escapes (#3742)
Co-authored-by: Zhengchao An <anzhengchao@gmail.com>
2026-06-22 19:28:40 +08:00
houseme 9b1272529a fix(ecstore): avoid blocking fs in async paths (#3586) 2026-06-19 01:44:17 +08:00
houseme e8012bd1ba refactor(logging): normalize admin telemetry and error messages (#3430) 2026-06-14 13:27:10 +08:00
houseme 82af181dcf refactor(logging): unify governance runtime events (#3367) 2026-06-11 22:26:02 +08:00
houseme b1339a98ed fix(scanner): delay deep verify for fresh objects (#3329)
Gate scanner-triggered deep heal behind a short new-object cooldown and retry transient bitrot shard size mismatches during local verification.

Add targeted tests for cooldown mode selection and size-mismatch classification, and log when deep scans are downgraded or size-mismatch retries occur.
2026-06-10 15:50:44 +00:00
houseme 3fed21c68a fix(startup): add disk init diagnostics (#3324)
* ci(build): remove macOS x86_64 target

* ci(build): reduce workflow timeout

* fix(startup): add disk init diagnostics
2026-06-10 14:24:12 +00:00
cxymds 61f0dfbc40 fix(ecstore): invalidate wiped disk id cache (#3251) 2026-06-07 04:25:38 +00:00
安正超 4d50c72e4c fix(ecstore): skip hidden metadata in walk limit (#3224)
* fix(ecstore): skip hidden metadata in walk limit

* fix(ecstore): match walk limit to visible versions

* fix(ecstore): avoid decoding all versions for limit
2026-06-05 13:50:02 +00:00
houseme f708c22b0e fix: make ecstore trash cleanup idempotent (#3205)
* fix(http): reduce header timeout log noise

Log peer_addr for HTTP connection errors so idle or probe traffic can be traced back to the source.

Downgrade HeaderTimeout connection logs from warn to info because these timeouts are often caused by incomplete probe traffic rather than a service-side fault.

* fix(ecstore): suppress spurious warn logs for missing trash cleanup sources

Extract `reliable_rename_inner` with a `warn_on_failure` flag so that
`rename_all_ignore_missing_source` can skip the warn log when the source
is absent. This completes the idempotent cleanup fix — the previous
implementation still emitted a `reliable_rename failed` warning before
the NotFound was caught and silenced.

Also restore the `recursive` branching in `move_to_trash` so that
non-recursive deletions use `fs::rename` directly (preserving original
semantics), while recursive deletions use the idempotent helper.
2026-06-04 04:21:31 +00:00
cxymds 39a283fbce fix(windows): handle zfs volume paths (#3157)
* fix(windows): handle zfs volume paths

* fix(windows): address volume path review

* refactor(windows): share fallback path resolution

---------

Co-authored-by: houseme <housemecn@gmail.com>
2026-06-01 13:04:40 +00:00
安正超 cf55743579 perf: reduce spawn_blocking contention in PUT path (#3132)
* perf: reduce spawn_blocking contention in PUT path (~23% throughput gain)

Flame graph profiling identified tokio blocking pool mutex contention
as the #1 bottleneck (17.3% of CPU time). Each spawn_blocking call must
acquire parking_lot::raw_mutex to enqueue work. With 16 concurrent PUTs
× 4 disks × 3+ spawn_blocking per disk, this became a serialization point.

Optimizations applied:
- Merge make_dir_all + file write into single spawn_blocking
- Merge read_file + parse + write + rename into single spawn_blocking
  for inline objects (small files)
- Optimize reliable_rename to try rename first, mkdir only on ENOENT
- Optimize remove/remove_std to try remove_file first, EISDIR fallback
- Add encode_inline_small fast path for small objects
- Parallelize bitrot writer creation with join_all

Benchmark (4KiB PUT, 4-disk EC, 16 concurrent, 8 rounds, randomized A/B):
  Baseline: ~950 obj/s → Optimized: ~1173 obj/s (+23%)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

* fix: address review comments

- Return old_data_dir for non-inline rename_data path (was incorrectly None)
- Restore delete_all cleanup of PUT temp data on failure paths
- Fix cargo fmt formatting

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

* style: apply rustfmt from stable 1.96.0

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

* fix: collapse nested if-let chains for clippy compliance

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

* fix: address Copilot review comments

- fs.rs: handle macOS EPERM from remove_file on directories
- os.rs: restore NotFound=Ok(()) semantics on first rename attempt
- local.rs: use try-rename-then-mkdir pattern for inline rename_data

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

* test: add unit tests for encode_inline_small fast path

* test: fix comment and line length in encode_inline_small tests

* fix: revert reliable_rename and write_all_internal to match original

Restore the original reliable_rename logic (check parent exists, then
rename in loop) and the original write_all_internal (make_dir_all outside
spawn_blocking). The optimization changes caused a CI-only test failure
in capacity_dirty_scope_test that could not be reproduced locally.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

* fix: update comment and fix formatting for CI

- Fix macOS/BSD comment to accurately say macOS only
- Fix encode_inline_small test formatting to match rustfmt 1.96.0

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

* fix: propagate inline rename errors

* fix: retry rename when parent missing

---------

Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
2026-05-31 01:28:48 +00:00
houseme 1de0ba916d fix(ecstore): reduce restart-time startup race noise (#3108)
Treat empty ping bodies as liveness probes, add a startup cleanup barrier for early walk_dir calls, and delay immediate background cleanup/capacity timers to reduce transient restart-time VolumeNotFound noise.

Also downgrade expected missing-path producer results during startup from generic errors to warnings while preserving existing storage semantics.
2026-05-28 15:14:30 +00:00
安正超 22bcfc474e perf(ecstore): use direct std writes for local disk (#3069)
* perf(ecstore): use direct std writes for local disk

* fix(ecstore): avoid blocking disk writes in local disk path

* fix(ecstore): use tokio async write in local disk path

* chore(ecstore): remove unused AsyncWriteExt import

* fix(ecstore): preserve write semantics and avoid bytes copy

* fix(ecstore): optimize write_all_internal buffer paths

* fix(ecstore): sync write path in local disk

* fix(ecstore): align sync flag docs and avoid ref copy

* chore(rust): format local write path

---------

Co-authored-by: houseme <housemecn@gmail.com>
2026-05-23 15:22:49 +00:00
houseme e69e32285b fix(ecstore): harden multipart part metadata visibility (#3042)
* fix(ecstore): harden multipart part metadata visibility

* fix(ecstore): address PR3042 review follow-ups
2026-05-21 14:47:52 +00:00
安正超 dd35538e90 fix(ecstore): remove stale disk TODOs (#3022) 2026-05-20 03:59:34 +00:00
安正超 a8a5f25af3 perf(ecstore): remove owned write sync regression (#3018) 2026-05-19 22:14:35 +00:00
houseme 25c6bdf490 perf(filemeta): phase-1~3 rename_data metadata optimization (#3011)
* chore(perf): harden amd64 profiling benchmark flow

* fix(profiling): isolate bench buckets and map protobuf conflict

* perf: avoid blocking owned local writes

* style: format profile admin handler

* docs: clarify observability trace validation

* perf: reduce mkdir overhead on local writes

* perf: add rename_data meta microbenchmark

* perf(filemeta): fast-path data_dir decode in version meta

* perf(filemeta): collapse data-dir lookup into one scan

* perf(filemeta): reduce scan allocs and refresh meta bench

* perf(ecstore): skip mkdir path on read-only open

* perf(filemeta): single-pass unshared data-dir scan

* perf(filemeta): add two-key inline remove fast path

* perf(filemeta): compare remove-two keys by bytes first

* bench(ecstore): add remove_two-only micro benchmark

* bench(ecstore): stabilize rename_data meta benchmark timing

* bench(ecstore): align rename_data path with remove_two

* perf(filemeta): avoid uuid string alloc in remove_two

* perf(filemeta): add fast-path for empty inline data

* perf(filemeta): streamline add_version match branch

* perf(filemeta): fast-return remove_key on miss

* perf(filemeta): speed up add_version insertion lookup

* style(ecstore): normalize formatting in perf-tuning files

* refactor(filemeta): unify inline data removal paths
2026-05-19 10:20:24 +00:00
安正超 a9e62dc2c2 perf: avoid blocking hop for owned disk writes (#3004)
* perf: avoid blocking hop for owned disk writes

* fix(ecstore): sync owned write path
2026-05-19 02:27:14 +00:00
wood 0e888cfef2 fix(ecstore): list_object_v2 error when scanning multipart folder (#2946)
* fix(ecstore): list_object_v2 error when scanning multipart directory with prefix

Signed-off-by: w0od <dingboning02@163.com>

* test(ecstore): cover prefix dir scan with multipart folder support

Signed-off-by: w0od <dingboning02@163.com>

* test(ecstore): harden test shape to improve regression detection

Signed-off-by: w0od <dingboning02@163.com>

* refactor(ecstore): move multipart dir filter into recursion to reduce I/O

Signed-off-by: w0od <dingboning02@163.com>

* test(ecstore): replace scan_dir test with walk_dir integration coverage

Signed-off-by: w0od <dingboning02@163.com>

* refactor(ecstore): remove unnecessary .clone() calls

Signed-off-by: w0od <dingboning02@163.com>

---------

Signed-off-by: w0od <dingboning02@163.com>
2026-05-18 11:01:20 +00:00
Henry Guo bca8b08c2b fix: handle Windows paths in pre-commit tests (#2974)
Co-authored-by: houseme <housemecn@gmail.com>
2026-05-15 14:04:51 +00:00
Sergei Z. 8c49671161 Fix #2775 recursive list handling in LocalDisk::scan_dir() (#2923)
Co-authored-by: houseme <housemecn@gmail.com>
Co-authored-by: 安正超 <anzhengchao@gmail.com>
2026-05-13 14:36:23 +00:00
weisd c5125b20bd fix(storage): keep storage info RPC map encoded (#2942)
Co-authored-by: cxymds <Cxymds@qq.com>
2026-05-13 06:19:43 +00:00
weisd 9b83de53e1 fix(ecstore): surface prefix listing storage errors (#2940) 2026-05-13 04:24:54 +00:00
安正超 4728abcff1 fix(security): document unsafe and TLS overrides (#2835) 2026-05-06 15:09:02 +00:00
安正超 50ddec3ffc fix: preserve data on self metadata copy (#2819)
Co-authored-by: cxymds <Cxymds@qq.com>
2026-05-06 06:28:36 +00:00
houseme 50d03ef021 perf(memory): add reclaim signals and cache controls (#2689) 2026-04-26 16:42:35 +00:00
houseme 81854762d4 fix(ecstore): log missing local paths during physical disk independence checks (#2680) 2026-04-25 08:01:01 +00:00
houseme 1e9c75a201 feat(ecstore): enforce local disk topology guardrails and expose device ids (#2679) 2026-04-25 06:02:02 +00:00
weisd f08b592c6f fix(storage): list prefix children behind marker objects (#2643) 2026-04-22 11:23:27 +00:00
Tunglies 579b124726 lint: clippy rules or_fun_call (#2561)
Co-authored-by: houseme <housemecn@gmail.com>
2026-04-16 02:48:39 +00:00
安正超 b2b92de26c test(ecstore): cover fresh tmp cleanup path (#2562) 2026-04-16 02:12:16 +00:00
weisd da9dfb51e7 fix(ecstore): cleanup stale put object temp data (#2529) 2026-04-15 14:40:58 +08:00
安正超 6d476ae9a1 revert: remove BlockReadable trait and chunk-based I/O from rio, ecstore, disk layers (#2533) 2026-04-14 19:09:02 +08:00
weisd a8a2aaa460 fix(ecstore): avoid duplicate keys in ListObjectsV2 (#2467) 2026-04-10 11:29:23 +08:00
houseme 32bf8f5bf3 feat(storage): add direct chunk GET fast path (#2351)
Signed-off-by: houseme <housemecn@gmail.com>
Co-authored-by: heihutu <heihutu@gmail.com>
Co-authored-by: cxymds <Cxymds@qq.com>
2026-04-07 08:33:46 +08:00
安正超 d5f05993a3 test(ecstore): cover read offset overflow (#2341) 2026-03-30 20:48:03 +08:00
houseme 7172e151de fix: address correctness, safety, and concurrency issues (#2327)
Co-authored-by: heihutu <heihutu@gmail.com>
Co-authored-by: houseme <housemecn@gmail.com>
2026-03-30 00:30:57 +08:00
weisd 2e7abfbd63 fix(ecstore): preserve raw metadata read semantics (#2258)
Signed-off-by: houseme <housemecn@gmail.com>
Co-authored-by: houseme <housemecn@gmail.com>
Co-authored-by: heihutu <heihutu@gmail.com>
2026-03-23 10:42:16 +08:00
weisd a42320848c fix(ecstore): invalidate xl.meta cache after writes and rename (#2255) 2026-03-22 11:30:29 +08:00
weisd b9b7d86ae4 feat: improve legacy metadata and admin compatibility (#2202) 2026-03-18 21:05:09 +08:00
evan slack d52a10c5fb chore(obs): Improve tracing instrumentation (#2086)
Co-authored-by: loverustfs <hello@rustfs.com>
2026-03-07 20:03:20 +08:00