VOI-68: Harden worker stale job handling #195

Merged

backmeupplz merged 1 commit from voi-68-harden-worker-job-not-found into main

2026-05-17 02:43:00 +00:00

backmeupplz commented

2026-05-17 02:38:00 +00:00

(Migrated from github.com)

Summary

Classify worker API 404 stale-job responses (job_not_found plus future ownership/state codes) separately from retryable network/server errors.
Stop stale local work items on source/downloaded/start/heartbeat/result/failure paths without reporting another failure or crashing the worker loop.
Add worker-client proof coverage for stale start, result upload, heartbeat, and failure-report responses, including queue continuation and sanitized logging.

Validation

yarn test:worker-client - passed
yarn build-ts - passed
yarn lint - passed

Manual QA: not required for this change; the behavior is covered by local HTTP proof servers simulating the worker API stale ownership races. Windows-host smoke was not run because no staged Windows worker deployment was needed for the client-side classification change.

## Summary - Classify worker API 404 stale-job responses (`job_not_found` plus future ownership/state codes) separately from retryable network/server errors. - Stop stale local work items on source/downloaded/start/heartbeat/result/failure paths without reporting another failure or crashing the worker loop. - Add worker-client proof coverage for stale start, result upload, heartbeat, and failure-report responses, including queue continuation and sanitized logging. ## Validation - `yarn test:worker-client` - passed - `yarn build-ts` - passed - `yarn lint` - passed Manual QA: not required for this change; the behavior is covered by local HTTP proof servers simulating the worker API stale ownership races. Windows-host smoke was not run because no staged Windows worker deployment was needed for the client-side classification change.