dheerajturaga opened a new pull request, #65830:
URL: https://github.com/apache/airflow/pull/65830
When an Edge worker drains while its API server is unreachable, the
supervisor subprocess's inherited asyncio signal wakeup fd caused
signals in the child to re-fire the parent's shutdown handler at
~9.5 kHz, flooding logs with tens of thousands of SIGTERM received
messages per second.
Reset the inherited signal state (wakeup fd, SIGTERM/SIGINT/SIG_STATUS
handlers) in the forked supervisor child before running supervise(),
and make shutdown_handler idempotent so any residual re-fires are
no-ops. Also drop the racy os.setpgid(child_pid, 0) call from the
handler — it was a no-op when the child had already called setpgrp()
and raised EPERM once the child had exec'd.
---
##### Was generative AI tooling used to co-author this PR?
<!--
If generative AI tooling has been used in the process of authoring this PR,
please
change below checkbox to `[X]` followed by the name of the tool, uncomment
the "Generated-by".
-->
- [ ] Yes (please specify the tool below)
--
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]