aglinxinyuan commented on PR #6797:
URL: https://github.com/apache/texera/pull/6797#issuecomment-5054601295

   Pushed 35203dc950 to fix a self-inflicted hang in this PR's own CI backstop.
   
   The `UDF_PYTHON_LOG_STREAMHANDLER_LEVEL=WARN` I added flows through 
`udf.conf` into every spawned Python UDF worker, which configures loguru with 
it — and loguru has no `WARN` level, only `WARNING`. So each worker died at 
startup with `ValueError: Level 'WARN' does not exist` before handshaking back 
to the JVM, and `PythonProxyClient` then blocked on the port promise 
(`Await.result`, no timeout) → `amber-integration` hung until the 6 h cap. Both 
the earlier commit and the head sat ~3 h in the integration step before I 
cancelled them. The `amber` unit job stayed green because `skip-integration` 
never spawns a worker, so `WARN` never reached loguru there.
   
   Fix:
   - `UDF_PYTHON_LOG_STREAMHANDLER_LEVEL=WARNING` for the loguru side; **keep** 
`TEXERA_SERVICE_LOG_LEVEL=WARN` for logback — that's its actual level name, and 
logback silently falls back to `DEBUG` on an unknown string, so `WARNING` there 
would flood logs (the opposite of the intent).
   - Added `timeout-minutes: 40` to `amber-integration` so a worker-startup 
deadlock fails fast instead of burning a full runner — no job in `build.yml` 
had a job-level timeout before.
   
   `amber-integration` is green again on the new run (ubuntu finished the 
integration step in ~8.5 min; macOS running normally).
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to