ci: full-suite (ci-full) validation for #23484 — BatchChonkVerifier handle teardown#23901
Closed
AztecBot wants to merge 8 commits into
Closed
ci: full-suite (ci-full) validation for #23484 — BatchChonkVerifier handle teardown#23901AztecBot wants to merge 8 commits into
AztecBot wants to merge 8 commits into
Conversation
Random-linear-combine (via a fresh challenge gamma) the G_0 binding check and the main IPA relation in IPA::batch_reduce_verify so the batch performs a single SRS MSM / IPA commitment instead of two.
Bound the graceful-stop phase (send-queue flush and the chonkBatchVerifierStop native RPC) with a timeout and move FIFO-reader/socket/native-process teardown into a finally block, so an unresponsive native backend can never leave the FIFO read stream (a blocking threadpool read that cannot be unref'd) or the unix socket open. This keeps the host process — e.g. the e2e prover-full test — exiting cleanly instead of hanging at teardown. Also unref the per-request result timeout so an orphaned request can't hold the event loop open.
AztecBot
added
ci-draft
…n exit The FIFO result reader used fs.createReadStream, which services the pipe with a blocking threadpool read. destroy() cannot cancel that read while a writer is still attached, so if the native bb verifier process outlives teardown (e.g. it has not yet exited after SIGTERM) the read stays parked and keeps the host process alive — the e2e prover-full test passes all assertions but Jest 'did not exit' and is eventually killed (exit 124). Open the FIFO O_NONBLOCK and read it through a libuv pipe handle (net.Socket) instead. Reads are then epoll-based, so stop()'s destroy() releases the handle immediately regardless of the writer, letting the process exit cleanly.
AztecBot
force-pushed
the
cb/babysit-23484-ci-full
branch
from
e1019c9 to
ae24e75
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
3 participants
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.
CI-only vehicle to run the ci-full suite (incl.
e2e_prover_full) against the exact head of #23484, because theci-fulllabel can't be applied to that PR from this session.Contains the full #23484 changeset plus the fix commit
0cc1020("ensure BatchChonkVerifier.stop always tears down handles"). The real fix lives on #23484 — do not merge this PR; it exists only to exercise prover-full under ci-full and confirm the prover-full hang is resolved.Ref: #23484
Created by claudebox · group:
slackbot