sigul

#28 Configure an alarm to kill requests that hang
Merged by jcline. Opened by jcline.
jcline/sigul timeout-requests  into  master

Download 28.patch

Register an alarm signal handler for the request handler child process
and configure an alarm for an hour per request. The alarm handler exits
the child process, which will cause a warning-level log before
continuing to handle requests.

Signed-off-by: Jeremy Cline jeremycline@linux.microsoft.com

Will that mean that particular stuck package signing will retry? Or silently just not have happened?

ie, right now if we kill things and restart, robosignatory just gets an error back via the bridge and puts that message back on the queue and then retrys.

I hope it means the bridge retries, but hope isn't a great strategy. I can spend some time next week reading through it to make sure that's right.

The way you've been unsticking things is by restarting the server process, correct? I think this should be roughly equivalent to that.

Yeah, restarting the server, it reconnects to the bridge and the bridge notices that it's reconnected and sends a 'failure' back to the client, and the client just retries the entire package.

rebased onto a3c76ae339670a309681c883771172c46409488a

rebased onto edb6f3ba1fc5d704364ec97dec9ef32580c2164f

Pull-Request has been merged by jcline
Metadata
None
No Tags
Static archive. pagure.io was sunset in 2026; this is a read-only snapshot.
Red Hat Logo

Fedora is sponsored by Red Hat. Learn more about the relationship between Red Hat and Fedora »

© Red Hat, Inc. and others.