Repository navigation
fix(dlq): skip requests owned by another live batch when failing a batch - #998
Open
burgulvedant wants to merge 1 commit into
Open
burgulvedant wants to merge 1 commit into
burgulvedant wants to merge 1 commit into
Conversation
The DLQ reconciler for batch stages failed every member request of a dead lettered batch, including a request that a different live batch had already enrolled. That batch can go on to land the change, leaving the request reported as error for a change that merges. Reuse owningBatch when failBatch walks the member requests, excluding the batch being reconciled so it does not count as its own members' owner. Closes uber#826
burgulvedant
requested review from
a team,
behinddwalls and
sbalabanov
as code owners
October 31, 2026 12:13
This branch has not been deployed
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Why?
When a lost ack leaves two batch attempts for one request, the one that dead letters can fail a request that the other, still live, batch carries. That batch can go on to land the change, so the request is reported as error for a change that merges.
What?
failBatch now calls owningBatch for each member request and skips any request that a different live batch owns. owningBatch takes an optional batch to exclude, so the batch being reconciled does not count as its own members' owner. The request controller passes an empty value and behaves as before.
Test Plan
New unit tests cover a request carried by another live batch (created, speculating, landing, cancelling), and a request that is still failed when no other live batch carries it (no other batch, or the other batch failed, succeeded or cancelled). Existing fan out tests gained a mock for the request batch store. Run with
go test ./submitqueue/orchestrator/controller/dlq/...ormake test.Issue
Closes #826