Hi,
On Wed, Sep 20, 2006 at 03:42:05PM +0100, Philip Hands wrote:
> Philip Hands wrote:
> [...]
> > Pierre,
> >
> > Perhaps you're aware of what it is that you're doing to use all those
> > files in /tmp on gluck (a.k.a. cvs.debian.org)? If so, please stop it, or
> > do it another way that doesn't break things.
> >
> > It looks to me as though you're firing up loads of cvs instances, and
> > then killing them in a way that prevents them from clearing up after
> > themselves, but that's only after a cursory examination -- you should have
> > a better idea of the details of what's occurring.
> >
> > Looking at the graphs, you'll have consumed all the inodes again in
> > about 5 hours. If you've not done something about it before then, I'll
> > almost certainly be forced to kill all your cvs processes, and discard all
> > the directories belonging to you in /tmp on gluck.
>
> OK, so it accelerated madly and ate all the inodes again as soon as I wrote
> that, so I've now killed the older half of your cvs server processes, and
> removed all the /tmp/cvs-serv* no longer associated with an active process.
> That should still leave you with enough debris to work out what's wrong
> with whatever you're doing. Oh, I've left /tmp/cvs-serv31300 as well, in
> case you need an example of one of the abandoned directories to understand
> what's happened.
>
> That should hold it for a few a day or so (we're back to 44% inodes used)
I had an ethernet outage on my gateway, that why there was so many
process. I talk with phil on irc this afternoon. I stopped my cron.
Cheers,
--
Pierre Machard
<pmachard@debian.org> http://debian.org
GPG: 1024D/23706F87 : B906 A53F 84E0 49B6 6CF7 82C2 B3A0 2D66 2370 6F87
Attachment:
signature.asc
Description: Digital signature