flake: TestSubscribeRelayEstablishedMidStream

[flake-investigator]

CI Run: https://github.com/coder/coder/actions/runs/24092786705
Failed Job: test-go-pg (ubuntu-latest)
Commit: 21c08a37d7ad9e872721835284e80597120b041a (author: Jake Howell)

Failure:

=== FAIL: enterprise/coderd/x/chatd TestSubscribeRelayEstablishedMidStream (25.10s)
    chatd_test.go:1516: timed out waiting for worker to start streaming

Error analysis:

• Single test failure in enterprise/coderd/x/chatd/chatd_test.go.
• Test waits for the worker to begin streaming after the relay is established; it timed out, indicating a timing-sensitive relay/worker coordination race.
• No panic, OOM, or data race indicators observed in the job logs.

Root cause classification: Flaky test (timing-sensitive relay/stream startup).

Assignment analysis:

• git blame -L 1415,1485 enterprise/coderd/x/chatd/chatd_test.go points to commit e18094825a7 ("fix: retain message_part buffer for cross-replica relay") by Kyle Carberry, which introduced TestSubscribeRelayEstablishedMidStream.
• Recent file history (git log --oneline -10 --follow enterprise/coderd/x/chatd/chatd_test.go) also shows e180948 as the most recent substantive change touching this test.

Assignee: @kylecarbs

Related issues search (coder/internal):

• "TestSubscribeRelayEstablishedMidStream"
• "timed out waiting for worker to start streaming"
• "TestSubscribeRelay" / "chatd_test.go" / "SubscribeRelay"
• Closed issue 1430 (TestSubscribeRelayReconnectsOnDrop) is a different test/error.

Reproduction (if needed):

go test ./enterprise/coderd/x/chatd -run TestSubscribeRelayEstablishedMidStream -count=1

[flake-investigator]

New occurrence (2026-04-09):

• CI Run: https://github.com/coder/coder/actions/runs/24172986083
• Failed Job: test-go-pg (macos-latest)
• Commit: d9544603804c7cd4f6505a6b226c12e9623c7261 (author: Matt Vollmer)

Failure:

=== FAIL: enterprise/coderd/x/chatd TestSubscribeRelayEstablishedMidStream (61.81s)
    chatd_test.go:1539: timed out waiting for worker to start streaming (chat status=waiting, last_error="")

Notes:

• Single-test failure; no panic/OOM/data race indicators in the job logs.
• Same timing-sensitive wait for the worker to begin streaming; still timing out under CI load (macOS runner). Root cause remains flaky test / relay-worker coordination timing.

[zedkipp]

Another one: https://github.com/coder/coder/actions/runs/24204941397/job/70658242851

Build logs:

job-logs.txt

Snippet:

=== Failed
=== FAIL: enterprise/coderd/x/chatd TestSubscribeRelayEstablishedMidStream (60.19s)
    t.go:111: 2026-04-09 17:54:58.196 [debu]  pubsub: pubsub dialing postgres  network=tcp  address=127.0.0.1:5432  timeout_ms=0
    t.go:111: 2026-04-09 17:54:58.196 [debu]  pubsub: pubsub postgres TCP connection established  network=tcp  address=127.0.0.1:5432  timeout_ms=0  elapsed_ms=0
    t.go:111: 2026-04-09 17:54:58.225 [debu]  pubsub: pubsub connected to postgres
    t.go:111: 2026-04-09 17:54:58.225 [debu]  pubsub: pubsub has started
    t.go:111: 2026-04-09 17:54:58.227 [debu]  pubsub: started listening to event channel  event=chat:config_change
    t.go:111: 2026-04-09 17:54:58.292 [debu]  pubsub: started listening to event channel  event=chat:stream:b46b750d-1722-4264-98c2-51c8d04414b4
    t.go:111: 2026-04-09 17:54:58.316 [debu]  pubsub: publish  event=chat:stream:b46b750d-1722-4264-98c2-51c8d04414b4  message_len=2
    t.go:111: 2026-04-09 17:54:58.320 [debu]  pubsub: publish  event=chat:stream:b46b750d-1722-4264-98c2-51c8d04414b4  message_len=20
    t.go:111: 2026-04-09 17:54:58.322 [debu]  pubsub: publish  event=chat:owner:8f971cd3-c21a-49eb-b229-ba5623a92988  message_len=485
    t.go:111: 2026-04-09 17:54:58.323 [info]  processor: processing chat request  chat_id=b46b750d-1722-4264-98c2-51c8d04414b4
    t.go:111: 2026-04-09 17:54:58.323 [debu]  pubsub: publish  event=chat:stream:b46b750d-1722-4264-98c2-51c8d04414b4  message_len=71
    t.go:111: 2026-04-09 17:54:58.326 [info]  processor: chat interrupted  chat_id=b46b750d-1722-4264-98c2-51c8d04414b4
    t.go:111: 2026-04-09 17:54:58.357 [debu]  pubsub: publish  event=chat:stream:b46b750d-1722-4264-98c2-51c8d04414b4  message_len=20
    t.go:111: 2026-04-09 17:54:58.363 [debu]  pubsub: publish  event=chat:owner:8f971cd3-c21a-49eb-b229-ba5623a92988  message_len=485
    chatd_test.go:1539: timed out waiting for worker to start streaming (chat status=waiting, last_error="")

[ethanndickson]

Got a fix for this in progress but it's kinda complicated