I think this a known issue as it occurs pretty often, but it would be nice if we could find a fix/improvement for this. As I see it, the steps listed on "startTalos.bat" file are executed each time. - one common situation when an XP tester gets stuck is when running "runslave.py" script. "Runslave.log" does not offer much useful info on this though (see attachments). - I've also seen cases where "startTalos.bat" file is not executed, so the machine won't do anything until rebooted The XP testers in such a state need manual reboot before taking any job (this is usually done by :philor and buildduty).
For the first one: http://superuser.com/questions/35698/how-to-supress-terminate-batch-job-y-n-confirmation/35741#35741 Do we still deploy these files via puppet? On which repo is startTalos.bat? Is this it? https://dxr.mozilla.org/build-central/source/puppet/modules/buildslave/files/startTalos-xp.bat Where's the code that generates runslave.log? Is this the code? https://dxr.mozilla.org/build-central/source/puppet/modules/buildslave/files/runslave.py What generates log_error.log?
(In reply to Armen Zambrano [:armenzg] (EDT/UTC-4) from comment #4) > Do we still deploy these files via puppet? HW Windows testers are managed by GPO, not puppet. > On which repo is startTalos.bat? > Is this it? > https://dxr.mozilla.org/build-central/source/puppet/modules/buildslave/files/ > startTalos-xp.bat Taking a look at how startTalos.bat looks like on an XP testers, it seems that it's not the same file as the one stored in puppet. I was able to find it on the winadmin machine (there's a GPO policy pointing to it), but I'm not sure about the deploy process. > Where's the code that generates runslave.log? > Is this the code? > https://dxr.mozilla.org/build-central/source/puppet/modules/buildslave/files/ > runslave.py Yes. The log file also contains stores the "echo" entries from startTalos.bat that correspond to the cleaning process. > What generates log_error.log? log_error.log seemss to contain the errors generated by the test run.
Bug 1088032 was the previous instance of this I was having trouble finding (bug 972867 being the easier one to find).
Windows XP backlog hasn't been an issue recently. Moreover, XP testing will be dropped starting with Firefox 53 (see bug 1310836). That being said, I don't think it's worth spending more time working on this.