Repository navigation
Workspace startup issue #7366
Description
It’s still possible if workspace starts before account for it to get stale.
Failing withRetryConnUntilSuccess with error cause: Error: getaddrinfo ENOTFOUND account
at GetAddrInfoReqWrap.onlookupall [as oncomplete] (node:dns:120:26) {
errno: -3008,
code: 'ENOTFOUND',
syscall: 'getaddrinfo',
hostname: 'account'
}
This error originated either by throwing inside of an async function without a catch block, or by rejecting a promise which was not handled with .catch(). The promise rejected with the reason:
TypeError: fetch failed
at node:internal/deps/undici/undici:13185:13
at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
at async workerHandshake (/usr/src/app/bundle.js:240:591660)
at async /usr/src/app/bundle.js:240:590457
at async N6t.start (/usr/src/app/bundle.js:453:440228)
Metadata
Metadata
Assignees
Labels
No labels
Activity
MichaelUray commentedon Oct 20, 2026
The workspace service now waits for the account service: in server/workspace-service/src/service.ts the worker handshake runs with withRetryConnUntilSuccess inside a try/catch loop that retries until it succeeds, so starting before the account service no longer causes an unhandled rejection. It was checked against the current Platform-Collective develop code. If maintainers agree this covers the report, please consider closing.
MichaelUray commentedon Oct 20, 2026
@lexiv0re — based on the check described above, this looks resolved in the current code. Could you please close this issue if you agree? If you think it is not fixed yet, please describe what still goes wrong (steps, expected vs. actual behaviour), so it can be looked at again.