You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
fix(fullstack): re-seed on a bound in tier-2 realtime recovery checks (#1078)
* fix(fullstack): re-seed on a bound in tier-2 realtime recovery checks
The 1s settle (#1074) was not sufficient: postgres_changes has no
backfill, and a single seed that lands while the re-subscribed channel's
server-side binding is still activating is never delivered — 3/3 CI runs
timed out at the recovery assertion while local runs (higher latency,
larger natural slack) always passed. seedNoteUntilDelivered re-seeds up
to 3 times with a 7s delivery assertion per attempt: recovery means
delivery resumes, and each attempt still asserts real end-to-end
delivery. Verified locally against the real project: 14/14 green.
* fix(fullstack): unblock ci gates for the realtime reseed branch
- regenerate the public interface snapshot (sha-only drift for
./open-dropdown from the hygiene sweep's anchor sync; no public
declaration changes)
- fullstack:cloudflare-config-check: grant --allow-read to the test
invocation; the new queue-name parity test reads the workflow file
---------
Co-authored-by: DevBot <devbot@openelement.dev>
0 commit comments