* feat(heal): incremental heal status cursors and typed overlap policy (HS-06)
Incremental results: every retained result item now carries a monotonic
sequence number. The status query accepts a client cursor (sinceSeq on
the admin wire, Option<u64> internally) and returns only newer items,
plus nextSeq (the next cursor) and minSeq (the oldest retained
sequence). A cursor that fell behind the 1024-item retention window is
flagged through the existing truncated signal together with minSeq so
the client can restart from it. Sequencing survives task completion:
the completion archive stores the seq-stamped window. None keeps the
exact legacy full-snapshot behavior, so existing clients see no change.
Typed overlap handling for admin starts: RUSTFS_HEAL_OVERLAP_POLICY
(merge default | minio_error). Under minio_error, an admin start whose
path overlaps an active or queued task rejects with typed
already-running / overlapping-paths admission reasons (surfaced through
reason_label in the admin error body, sharing the existing
OperationAborted site because the s3s footprint ratchet forbids new
s3_error! sites); an exact duplicate start rejects with
already-running instead of silently merging. Scanner/autoheal/
read-repair sources never take the rejection path.
forceStart semantics now match MinIO for admin requests: an admin
forceStart first cancels the overlapping active admin task, then
admits the replacement.
Wire: the heal-control Query command grows an optional sinceSeq
(defaulted and skipped when absent, so older peers stay compatible);
the admin handler accepts the sinceSeq query parameter; the local
channel query gains the same cursor.
Tests: seq monotonicity and incremental slicing, window slide moving
minSeq with lagging-cursor flags, overlap matrix (same/containing/
contained/disjoint x policy x source), forceStart cancel-then-admit,
and the completion-archive window handoff.
Co-Authored-By: heihutu <heihutu@gmail.com>
* style: fmt after main merge
---------
Co-authored-by: heihutu <heihutu@gmail.com>
Co-authored-by: zhi22915 <qiuzgang@gmail.com>
* feat(common): add MRF intent channel and Mrf request source (HS-01)
Introduce the producer-facing half of the mission repair feed: a global
bounded (8192) channel carrying lightweight MrfIntent values from IO
error paths, plus the RUSTFS_HEAL_MRF_ENABLE delivery kill-switch and
config constants for queue/journal sizing. Delivery is strictly
non-blocking (try_send, drop-on-full) so it can sit on decode-failure
and partial-write paths without adding latency. HealRequestSource grows
a 'mrf' variant so admission accounting can attribute replayed intents.
Part of backlog#1865 (option a: wire HealEvent-style intents with a
durable retry ledger).
Co-Authored-By: heihutu <heihutu@gmail.com>
* feat(heal): add MRF queue, durable journal, and intent consumer (HS-01)
Consumer half of the mission repair feed: a bounded pending queue
(100k intents / 8 MiB dual ceiling, drop-newest on overflow), a durable
journal at buckets/.heal/mrf/journal.bin holding the unaccepted pending
snapshot, and a consumer task that batches intents off the global
channel, translates them into prioritized heal requests (decode
failure -> Urgent ECDecode, metadata corruption -> High Metadata,
partial write -> Normal object heal), and retries full admissions with
a 5s backoff and a 3-attempt ceiling.
Durability: every journal record carries its own CRC32 and a
format/version header, so a torn tail truncates cleanly at replay; the
journal is deleted after a successful replay and when the pending set
drains (mirroring MinIO's post-replay list.bin unlink). Losing the last
500 ms flush window is acceptable: replayed duplicates merge via the
manager dedup key and read-repair remains the safety net.
Metrics: rustfs_heal_mrf_queue_depth/_queue_bytes, _dropped_total
{reason}, _replayed_total, _journal_bytes, _journal_fsync_total.
The consumer is wired at heal runtime bootstrap right after manager
start, honoring RUSTFS_HEAL_MRF_ENABLE (default on, rollback = off).
Tests: unit tests for the dual ceiling, record roundtrip, torn-tail
truncation, and the priority mapping; integration tests against a real
4-disk ECStore proving channel intents reach the manager queue as
Urgent/mrf-attributed requests and journal replay arms intents, drops
torn tails, and removes the file.
Part of backlog#1865 (option a).
Co-Authored-By: heihutu <heihutu@gmail.com>
* feat(ecstore,scanner): deliver MRF intents from error paths (HS-01)
Wire the three production delivery points, each a single non-blocking
try_send next to the existing in-memory heal paths, which stay as the
fast path:
- read.rs decode-error branch: DecodeFailure intent beside the existing
read-repair submit, so an Urgent ECDecode request survives restarts
even when the Low-priority read-repair request was dropped or lost.
- add_partial: PartialWrite intent, giving partial-write recovery a
durable Normal-priority object heal across restarts.
- scanner_folder metadata-corruption classification: MetadataCorruption
intent beside the existing High-priority scanner heal request.
All three are on error paths only: zero cost on healthy IO.
Part of backlog#1865 (option a).
Co-Authored-By: heihutu <heihutu@gmail.com>
* fix: include mrf heal source counts
Co-Authored-By: heihutu <heihutu@gmail.com>
* fix: keep node heal status wire compatibility
Co-Authored-By: heihutu <heihutu@gmail.com>
---------
Co-authored-by: heihutu <heihutu@gmail.com>
* fix: resolve logic errors in ahm heal module
- Fix response publishing logic in HealChannelProcessor to properly handle errors
- Fix negative index handling in DiskStatusChange event to fail fast instead of silently converting to 0
- Enhance timeout control in heal_erasure_set Step 3 loop to immediately respond to cancellation/timeout
- Add proper error propagation for task cancellation and timeout in bucket healing loop
* fix: stabilize performance impact measurement test
- Increase measurement count from 3 to 5 runs for better stability
- Increase workload from 5000 to 10000 operations for more accurate timing
- Use median of 5 measurements instead of single measurement
- Ensure with_scanner duration is at least baseline to avoid negative overhead
- Increase wait time for scanner state stabilization
* wip
* Update crates/ahm/src/heal/channel.rs
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* refactor: remove redundant ok_or_else + expect in event.rs
Replace redundant ok_or_else() + expect() pattern with
unwrap_or_else() + panic!() to avoid creating unnecessary Error
type when the value will panic anyway. This also defers error
message formatting until the error actually occurs.
* Update crates/ahm/src/heal/task.rs
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* fix(ahm): fix logic errors and add unit tests
- Fix panic in HealEvent::to_heal_request for invalid indices
- Replace unwrap() calls with proper error handling in resume.rs
- Fix race conditions and timeout calculation in task.rs
- Fix semaphore acquisition error handling in erasure_healer.rs
- Improve error message for large objects in storage.rs
- Add comprehensive unit tests for progress, event, and channel modules
- Fix clippy warning: move test module to end of file in heal_channel.rs
* style: apply cargo fmt formatting
* refactor(ahm): address copilot review suggestions
- Add comment to check_control_flags explaining why return value is discarded
- Fix hardcoded median index in performance test using constant and dynamic calculation
---------
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
- Fix multi-threaded test conflicts in AHM heal integration tests
- Remove global environment sharing to prevent test state pollution
- Fix test_all_disk_method by clearing global disk map before test
- Improve data scanner and cache value implementations
- Update dependencies and resolve clippy warnings
Signed-off-by: junxiang Mu <1948535941@qq.com>
- Add global unbounded channel in common crate for heal requests
- Implement channel processor in ahm to handle heal commands
- Add Start/Query/Cancel commands support via channel
- Integrate heal manager initialization in main.rs
- Replace direct MRF calls with channel-based heal requests in ecstore
- Support advanced heal options including pool_index and set_index
- Enable admin handlers to send heal requests via channel
houseme