Repository navigation
Conversation
ApprovabilityVerdict: Approved at Macroscope's review found this PR approvable — This is a focused relay bug fix that keeps the existing catch-up worker alive for explicit relink or preference-change requests, without introducing idle polling or schema changes. A targeted regression test covers repeated catch-up and credential updates. Notes:
You can add or adjust custom eligibility rules. Learn more. |
|
Warning Review limit reachedOnly developers with an assigned seat can use this organization's usage-based review budget, and seats here are assigned manually. Ask an admin to assign a seat, or change the review continuation mode in Billing. Next included review available in 45 minutes. View limit details
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info
📝 Walkthrough
Merge Risk: ⚪ Minimal · up to
|
Dismissing prior approval to re-evaluate e0108ba
|
Closing in favour of #25201, which makes the same change. Thanks for the work here! |
After the startup activity snapshot succeeds, the relay catch-up consumer exits. Later requests caused by relinking or changing publication preferences remain queued, so active threads are not republished until another thread event occurs.
Keep the consumer alive and parked on its request queue after a successful snapshot. Explicit catch-up requests resume it; existing retry backoff stays in place and successful idle environments do not poll.
Verification: the added regression fails before the fix (one snapshot instead of two). All 33 AgentAwarenessRelay tests pass with the fix, including re-enabling publication, credential changes, and ten minutes of virtual idle with no extra snapshots or secret reads. Targeted lint and format pass. No live relay was used; the test proves consumer lifetime through the service harness.
Model: GPT-6.1-sol. Harness: Codex in T3 Code.