Bill Janssen <jans...@parc.com> writes: > David Bolen <db3l....@gmail.com> wrote: > >> Bill Janssen <jans...@parc.com> writes: >> >> > I'm trying to get a new buildbot in the swim of things, and it keeps >> > getting into this state where the buildslave process seems caught in an >> > endless loop. Perhaps someone here knows why? >> >> Do you have any information as to what it is doing while in the loop? > > Not much. The twistd log shows that it is, in fact, trying to reconnect > to the build master. It's just chewing up 100% CPU. Could just be > updates to the master that are doing this, but my other (PPC) buildbots > seem OK with it.
When I look at the slave status page[*], it seems to show all of your slaves as idle, so if one of them is looping trying to connect, I suspect you may have run into that last case I've seen (mismatch between slave and master). Try shutting down the build slave for a while and watching the status page until it shows disconnected. Then you should be able to restart things and have it attach successfully. Leaving the slave continually trying to reattach, I believe, prevents the master from ever declaring it disconnected. I'm guessing the master is rejecting the new connection since it thinks the slave is still attached (though the request seems to defeat the status from correcting). I'm not sure how it gets into that state but it's happened a handful of times to me (typically after some network interruptions, so maybe it just needs enough time for buildbot's own keepalives to notice what is otherwise a quiet TCP channel). There used to be a way to request a "ping" from the master side (I think on the same page you could manually run a build from) that I would used to force it to recognize a slave was really down, but after the web interface got rearranged a while back, I can't seem to find the option any more. -- David [*] http://www.python.org/dev/buildbot/buildslaves _______________________________________________ Python-Dev mailing list Python-Dev@python.org http://mail.python.org/mailman/listinfo/python-dev Unsubscribe: http://mail.python.org/mailman/options/python-dev/archive%40mail-archive.com
