Workstation 23 (Beta) locking up

Kamil Paral kparal at redhat.com
Thu Oct 8 12:08:49 UTC 2015


> Thanks for all suggestions, I tried to apply them and this is what I got.
> While I don't use the Android Studio I did often use Eclipse and other
> Java heavy processes in the past days (I'm a Java developer playing
> with big data, so the heavy apps idea sounded interesting).
> 
> So today I've been leaving the laptop on, but without using it at all:
> just logged in Gnome 3, no apps running.
> 
> After some hours it crashed, and (after waiting a bit and then
> rebooting), this is the output of `journalctl -b -1` after restarting
> it:
> 
> Oct 06 22:33:06 t440.slug rtkit-daemon[840]: The canary thread is
> apparently starving. Taking action.
> Oct 06 22:33:06 t440.slug rtkit-daemon[840]: Demoting known real-time
> threads.
> Oct 06 22:33:06 t440.slug rtkit-daemon[840]: Successfully demoted
> thread 5579 of process 5564 (/usr/bin/pulseaudio).

Well this looks like a problem somewhere, but probably not related to system freeze. You're interested at the very end of the log, if there's anything that will help you uncover why it froze.

You don't need to wait for another crash, just display the particular log again, using correct `-b <number` argument (in case you rebooted in the meantime). See `man journalctl` and see "-b" and "--list-boots".


> I then repeated the cycle, but with an open SSH terminal from my other
> workstation (Fedora 22).
> Again after some hours the system seemed locked up again. But the SSH
> terminal was not disconnected!
> I eventually got the "Broken Pipe" message, terminating the SSH
> session, only after I forced it off. So I guess it was not entirely
> dead?

It takes a long time before ssh realizes the connection is terminated. You can try to ping the machine to figure out whether it's still responding to network requests. If it doesn't, the ssh connection is probably just hanging. Again, you're interested in the last messages that were printed out before the freeze happened.

> 
> Is this information enough for you to guess something, or should I go
> ahead and try the F24 kernels?

Use fpaste to upload the last 50 or so lines (whatever looks suspicious) of the system journal and link it here. And sure, go ahead and try F24 kernels as well (or more recent F23 kernels, if there are any).


More information about the test mailing list