is there a better solution than "killall -9 journalctl"?

[Robert P. J. Day]

On Sun, 13 Jul 2014, Adam Williamson wrote:

> On Sun, 2014-07-13 at 22:28 -0400, Robert P. J. Day wrote:
> > On Sun, 13 Jul 2014, Adam Williamson wrote:
> >
> > > On Sun, 2014-07-13 at 15:23 -0600, Kevin Fenzi wrote:
> > > > On Sun, 13 Jul 2014 17:19:04 -0400 (EDT)
> > > > "Robert P. J. Day" <rpjday at crashcourse.ca> wrote:
> > > >
> > > > >
> > > > >   i'm still trying to understand why fedora rawhide thinks it's
> > > > > necessary to run a "journalctl" command on every one of my eight
> > > > > hyperthreaded cores. i'm so disgusted with this indescribable
> > > > > stupidity that i just gave up and am now running
> > > > >
> > > > >   $ sudo killall -9 journalctl
> > > > >
> > > > > every 10 seconds on /dev/tty2. i'd love a more elegant solution but,
> > > > > frankly, this is solving the problem so i'm sticking with it.
> > > >
> > > > When I saw this here it was due to abrt...
> > > >
> > > > systemctl restart abrtd
> > > >
> > > > caused it to stop doing anything here and it's not reappeared yet.
> > > > YMMV.
> > >
> > > Has someone who's seeing this filed it as a bug against abrt yet?
> >
> >   hmmmmm ... i tried that and it didn't seem to solve the problem. i'm
> > still reduced to the ugly solution i mention above -- killing
> > journalctl every 10 seconds. if i don't, multiple invocations of
> > journalctl eventually just take over my machine.
> >
> >   i'm sure there's a more elegant fix.
>
> well, I mean, the 'elegant fix' would be to figure out what's
> causing it to run excessively and stop that thing from doing that.
> It's not a given that there's any more elegant *workaround*, no.
>
> When one of these journalctl processes is running, 'systemctl status
> (pid of that process)' may tell you what service is spawning it. --

  i'll set aside some time this evening to look at this, but i just
want to confirm that i'm not the only person seeing this -- that is,
i have a quad core i7 with hyperthreading so i have 8 possible CPUs
and, without messing with any kind of related configuration, i've seen
this system with 7 journalctl processes running, each one pegged at
98-99% CPU. and that's more than enough to have the fans on this thing
running full blast to keep it from melting down. :-(

  and doing a "systemctl restart abrtd" didn't appear to solve the
problem.  more this evening after some testing ...

rday

-- 

========================================================================
Robert P. J. Day                                 Ottawa, Ontario, CANADA
                        http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================