Repository navigation
Conversation
This comment has been minimized.
This comment has been minimized.
b7r6
force-pushed
the
b7r6/fastslow-inflight-waiter-fresh-upload
branch
2 times, most recently
from
October 7, 2026 20:28
9c06b74 to
11d0614
Compare
… entry The inflight-dedup drop-guard in CacheLookupScheduler::add_action was bound as `let _scope_guard = scope_guard;`, so it fired on *every* exit of the spawn — including normal completion, after the spawn had already removed its own entry and resolved the coalesced waiters. A concurrent request for the same cacheable key, arriving in the window between the leader's own `inflight_cache_checks.remove(...)` and the spawn's exit (dropping the guard), inserts a fresh entry. The stale guard then removes *that* entry, orphaning the second request: its oneshot sender is dropped, surfacing a spurious "tx hung up" error to the client and forcing a needless retry. Fix: defuse the guard via `ScopeGuard::into_inner` the moment the cache check returns. From that point every arm removes our own entry with no intervening `.await` (no drop point), so defusing is leak-free and the guard can no longer clobber a concurrent same-key re-insertion. The guard still provides true drop-safety for the only window that matters: a spawn death during the cache-store check, before we claim the entry. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…es the digest An existence check that found a slow-store write in flight parked on that specific writer's completion mutex. A ByteStream upload that is interrupted (cancel, half-close, client deadline) is parked for resumption, keeping its FastSlowStore::update() future -- and thus the locked mutex -- alive. A later registration for the same key overwrote the map entry, so completing the digest under a fresh upload id never woke the original waiters: they starved until their client deadline even though the blob was present and new lookups succeeded. Replace the per-writer mutex with a per-key slot shared by all concurrent writers of that key, broadcasting Pending/Done over a watch channel. Any writer completing the digest wakes every waiter; the entry is removed (and waiters released with "not written") only when the last writer deregisters, which also fixes the stale-remove where a second writer's guard evicted a slot the first writer still occupied. Regression: stalled_upload_waiter_released_by_fresh_upload_of_same_digest -- fails on v1.7.3 (waiter hangs past a 2s deadline), passes with the fix; also asserts a genuinely-absent digest still resolves promptly. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
b7r6
force-pushed
the
b7r6/fastslow-inflight-waiter-fresh-upload
branch
from
October 8, 2026 17:51
11d0614 to
77dc359
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What and why
Lane A stability fix from the findings catalog (see
nativelink-findings-2026-10-05.csvon the
meta/findings-catalog-2026-10-05branch), rebased onto OSS main@bf4b3aa855. This fork PR exists to validate the change on CI that is
representative of upstream PR lanes (public read-only cache, same endpoints)
before any upstream submission.
How was this verified?
The branch carries its own regression tests (fails-without where the fix
class allows it — see the branch's test commits). This PR's purpose is the
full representative CI run; green here rehearses the exact checks an
upstream PR would face.
Risk
Fork-internal validation PR; nothing lands anywhere by merging this. The
underlying fix's risk assessment happens at upstream-submission time.
AI assistance
Agent-assisted pipeline (Claude) for rebase and PR mechanics under my
direction; the fix itself carries verification in its tests.