[indent]

On Nov 4 between 07:50 and 08:20 UTC, five stable accounts on different clusters got HTTP 404 for their Dot room for a few seconds, and the primary-Dot lookup also said there was no Dot:

warn The linked Dot room returned HTTP 404 {"primary_dot_missing":true,"primary_dot_replaced":false}

All five went BAD_CREDENTIALS chatgpt-dot-unavailable and logged nothing more. They only came back about 3 hours later, when a pod restart reconnected them to the same Dot room without the user signing in again. So ChatGPT's answer was temporary, but run treated it as final.

run returned on errDotUnavailable in both the connect loop and the poll loop. Now it's split into connectDot and serve. A missing Dot still sends chatgpt-dot-unavailable and stops polling and the live-signal stream. The client then re-verifies every 5 minutes, and once the Dot room answers again it resumes and goes back to CONNECTED. A Dot that is really gone or replaced keeps the same state as before, and signing in again works the same way.

go build ./..., go vet ./pkg/connector/ and go test ./pkg/... pass.

Tag @indent to continue the conversation here.

[indent]

Issues

All clear! No issues remaining. 🎉

1 issue already resolved

• When the Dot goes missing during profile refresh, serve waits 5 minutes before its deferred stream shutdown runs. The live-signal stream keeps running for the whole recheck interval, although the PR intends it to stop. (fixed by commit 4012459)
  _{Found by Indent Review Agent}

CI Checks

All CI checks passed on 4012459.

Review agents

Select any unchecked box below to run or rerun that agent.

Passed (1)

• Indent Review Agent · No issues found

Full results

Indent Review Agent

• Summary: No issues found
• Last ran on commit: 40124598
• Latest result

  {
    "summary": "No issues found",
    "findings": []
  }