fix(op-acceptance-tests): stabilize TestUnsafeGapFillAfterUnsafeReorg_RestartL2CL#20199
Merged
fix(op-acceptance-tests): stabilize TestUnsafeGapFillAfterUnsafeReorg_RestartL2CL#20199
Conversation
ajsutton
force-pushed
the
aj/fix/unsafe-gap-fill-reorg-flake
branch
2 times, most recently
from
637551e to
4a8ada2
Compare
…_RestartL2CL ReorgTriggered only waits until the block at the reorg-target height has been rewritten on the sequencer; it does not wait for the sequencer to rebuild past the verifier's frozen unsafe head. The test then asserts seq.Number > ver.Number, which races against the sequencer producing the next block after the reorg. In failing CI (op-geth, pipeline 123064, job 4866963) both heads were observed at 18 because the sequencer had reorged and rebuilt exactly up to the verifier's frozen height but had not yet extended past it. Capture the verifier's unsafe head immediately after stopping L2CLB (at which point the verifier cannot advance — its CL is down and the preset has NoDiscovery), then after ReorgTriggered returns wait for the sequencer to reach verFrozen.Number + 1 before sampling the heads for the assertion. This makes the seq > ver ordering structurally guaranteed rather than racing with post-reorg block production. Refs: #19936
…afeGapFillAfterUnsafeReorg_RestartL2CL The two pre-reorg Matched calls used only 5 attempts (~10s). This assumes the verifier is already in lockstep with the sequencer, which breaks when the verifier starts late relative to the sequencer and has to catch up tens of blocks — e.g. slow op-reth cold start or a gossipsub mesh that takes time to form under CI contention. When this happens the sequencer produces a burst of blocks to catch up to wall-clock genesis time, leaving the verifier well behind. 10s is not enough to close that gap via gossip. Bump both Matched budgets to 30 attempts (60s) — consistent with the existing post-stop gap-close budget of 50 attempts — so the verifier can finish catching up before the test perturbs the system. The happy path still returns on first success, so fast runs are unaffected.
ajsutton
force-pushed
the
aj/fix/unsafe-gap-fill-reorg-flake
branch
from
4a8ada2 to
3d8c640
Compare
wwared
approved these changes
Apr 22, 2026
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
2 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.
Claude: Opened by Claude on behalf of @ajsutton. Closes #19936.
Fixes two independent races in
TestUnsafeGapFillAfterUnsafeReorg_RestartL2CL. First,ReorgTriggeredreturns as soon as the reorg block is rewritten and does not wait for the sequencer to rebuild past the verifier's frozen unsafe head, so the immediately-followingseqUnsafe > verUnsafeassertion could sample the sequencer mid-rebuild and fail (matching the CI evidence ofseq:18 ver:18). Fixed by capturing the verifier's unsafe head right afterL2CLB.Stop()and waiting onReached(eth.Unsafe, verUnsafeFrozen.Number+1, 30)before the assertion, making the ordering structurally guaranteed.Second, the two pre-reorg
Matchedcalls used a 5-attempt (~10s) budget that assumed the verifier was in lockstep with the sequencer. Under cold-start contention (e.g. op-reth JIT compile) the sequencer can burst-produce 100+ blocks before the verifier begins syncing, and 10s isn't enough to close the gap. Bumped both budgets to 30 attempts (60s), consistent with the existing post-stop budget of 50.