feat(orchestrator): migrate the buildsignal poll loop to the hold primitive - #489
Draft
behinddwalls wants to merge 1 commit into
Draft
feat(orchestrator): migrate the buildsignal poll loop to the hold primitive#489behinddwalls wants to merge 1 commit into
behinddwalls wants to merge 1 commit into
Conversation
…mitive ## Summary ### Why? The orchestrator's buildsignal stage carried the same ack-and-republish poll loop the stovepipe stages just migrated off (previous commit): each non-terminal poll acked the delivery and PublishAfter'd the build id back to its own topic. Notably its republish reused the build id as the message id — the exact dedup collision that stalled stovepipe's loop (#465) was latent here — and a transient publish failure was wrapped retryable as the loop's only liveness, the awkward classification #469 removed on the stovepipe side. ### What? A non-terminal status now records a hold for the per-status poll delay and returns success; the framework postpones the delivery, which redelivers without counting toward the retry limit. publishBuild is deleted (the re-poll was its only caller); the speculate publish and halted-batch short-circuit are unchanged, and the DLQ reconciler is unaffected. The Process retryability comment is rewritten: the loop's continuation is framework-owned, so no publish needs a retryable wrap. build-runner.md's "Polling primitive" section is updated — hold supersedes the PublishAfter design it argued for, keeping the same retry_count semantics with no publisher, no minted ids, and no per-tick rows. ## Test Plan ✅ `bazel test //submitqueue/...` — NonTerminal cases assert Hold(per-status delay); Terminal/StatusError/UpdateStatusError/Halted cases fail on any Hold; the RepublishError test is deleted (a hold cannot fail). ✅ `make fmt`.
This was referenced Aug 1, 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
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.
Summary
Why?
The orchestrator's buildsignal stage carried the same ack-and-republish poll loop the stovepipe stages just migrated off (previous commit): each non-terminal poll acked the delivery and PublishAfter'd the build id back to its own topic. Notably its republish reused the build id as the message id — the exact dedup collision that stalled stovepipe's loop (#465) was latent here — and a transient publish failure was wrapped retryable as the loop's only liveness, the awkward classification #469 removed on the stovepipe side.
What?
A non-terminal status now records a hold for the per-status poll delay and returns success; the framework postpones the delivery, which redelivers without counting toward the retry limit. publishBuild is deleted (the re-poll was its only caller); the speculate publish and halted-batch short-circuit are unchanged, and the DLQ reconciler is unaffected. The Process retryability comment is rewritten: the loop's continuation is framework-owned, so no publish needs a retryable wrap. build-runner.md's "Polling primitive" section is updated — hold supersedes the PublishAfter design it argued for, keeping the same retry_count semantics with no publisher, no minted ids, and no per-tick rows.
Test Plan
✅
bazel test //submitqueue/...— NonTerminal cases assert Hold(per-status delay); Terminal/StatusError/UpdateStatusError/Halted cases fail on any Hold; the RepublishError test is deleted (a hold cannot fail). ✅make fmt.Stack