gsadmin/rustfs
1
0
Fork 0
mirror of https://github.com/rustfs/rustfs.git synced 2026-09-03 02:38:12 +00:00
Code Issues Packages Projects Releases Wiki Activity

perf(heal,scanner): single-flight MRF producers per detection event (#6282)

Browse Source 
Two producer paths double-booked the same damage across repair records
(backlog#1894 axis A):

- The scanner's corrupt-metadata branch fired a durable MRF journal
  intent, an immediate High heal request, and a pending-ledger entry for
  the same object. When the MRF intent is accepted into the channel it
  already covers the repair durably (the consumer files a High Metadata
  heal and the journal replays it across restarts), so the immediate
  request and ledger entry are dropped in that case; on delivery failure
  (feature disabled, channel uninitialized, or full) the old immediate
  request + ledger path runs unchanged, keeping the repair safety net.
- The read path filed a journal intent before the read-repair
  reservation check, so a burst of reads failing on one object booked a
  journal record per retry. The intent now rides the submission: it is
  filed only when the sighting wins the dedup TTL, next to the Low
  request, via a new optional mrf_intent field on
  ReadRepairHealSubmission (None keeps the historical no-intent
  behavior for the other read-repair call sites).

Manager dedup-key semantics are untouched; the fix is that competing
producers stop double-booking. With RUSTFS_HEAL_MRF_ENABLE off both
paths behave exactly as before.

Co-authored-by: heihutu <heihutu@gmail.com>
This commit is contained in:
houseme
2026-08-20 01:32:16 +08:00
committed by GitHub
parent d6efb65588
commit b1b4e443b2
4 changed files with 153 additions and 38 deletions
Download Patch File Download Diff File Expand all files Collapse all files
crates/ecstore/src/set_disk/read.rs
+20 -17
View File
@@ -1077,23 +1077,23 @@ impl SetDisks {
"Recoverable decode error triggered read repair"
);
let version_id = fi.version_id.as_ref().map(ToString::to_string);
// MRF journal intent: keeps a durable Urgent ECDecode
// request alive across restarts even when the in-memory
// read-repair request is dropped or lost (HS-01).
rustfs_common::mrf_channel::try_send_mrf_intent(
rustfs_common::mrf_channel::MrfKind::DecodeFailure,
bucket,
object,
fi.version_id,
);
submit_read_repair_heal(
bucket,
object,
version_id.as_deref(),
pool_index,
set_index,
Some(part_number),
"decode_error",
// Single-flight (backlog#1894 axis A): the durable
// MRF intent (Urgent ECDecode across restarts, HS-01)
// is bound to the read-repair reservation, so only the
// first sighting within the dedup TTL books a journal
// record instead of one per retried read.
submit_read_repair_heal_with_submitter(
ReadRepairHealSubmission {
bucket,
object,
version_id: version_id.as_deref(),
pool_index,
set_index,
part_number: Some(part_number),
reason: "decode_error",
mrf_intent: Some((rustfs_common::mrf_channel::MrfKind::DecodeFailure, fi.version_id)),
},
send_read_repair_heal_request,
)
.await;
has_err = false;
@@ -2577,6 +2577,7 @@ mod metadata_cache_tests {
set_index: 0,
part_number: Some(1),
reason: "missing_shards",
mrf_intent: None,
},
slow_read_repair_submitter,
)
@@ -2611,6 +2612,7 @@ mod metadata_cache_tests {
set_index: 0,
part_number: Some(1),
reason: "missing_shards",
mrf_intent: None,
},
dropped_read_repair_submitter,
)
@@ -2647,6 +2649,7 @@ mod metadata_cache_tests {
set_index: 0,
part_number: Some(1),
reason: "missing_shards",
mrf_intent: None,
},
capture_read_repair_submitter,
)