1114807 - runner fails when twistd.pid exists but is empty

Status: RESOLVED FIXED

(Release Engineering :: General, defect)

Description

[Chris Cooper [:coop] (he/him)]

I'm not sure how we got into this state on a slave (specifically talos-linux64-ix-007, but likely other slaves as well), but when the twistd.pid file exists but is empty, i.e. doesn't actually contain the process number, runner hits the following exception:

File "/opt/runner/tasks.d/4-buildbot.py", line 74, in <module>
main()
File "/opt/runner/tasks.d/4-buildbot.py", line 43, in main
if buildbot_process_id():
File "/opt/runner/tasks.d/4-buildbot.py", line 30, in buildbot_process_id
pid = int(f.read())
ValueError: invalid literal for int() with base 10: ''

I think the safest thing to do in this case would be to remove the twistd.pid file and *then* reboot, unless we have some other way tell if buildbot is already running.

Comment 1

[Morgan Phillips [:mrrrgn]]

Attachment: runner.bbtask.pidfile.diff

This is fixing up the issue on my testers.

Comment 2

[Morgan Phillips [:mrrrgn]]

Attachment: runner.bbtask.pidfile.diff

There was a small boo-boo on the last patch (cast to integer was left outside of the try statement).