On Tue, Feb 25, 2014 at 01:10:38PM +0000, Darac Marjal wrote:
> On Mon, Feb 24, 2014 at 01:57:02PM +0100, Tazman Deville wrote:
> > I have a little server running here in my office,
> > and logrotate kept running at c. 7am, and using up 100% CPU.
>
> Logrotate *itself* shouldn't use much CPU. But there are a couple of
> things I can think that might make it do so:
> * A badly-behaved {pre,post}rotate script. These would, I suspect, show
> up as separate processes, though.
How would one diagnose this? I'm having a similar problem,
but I haven't altered post/pre scripts, only the /etc/logrotate.d/
scripts for specific logs.
> * Compression of a large and/or corrupt log file.
This shouldn't happen on a regular basis, though, should it?
Unless something is routinely writing huge and/or corrupt logs,
but if they're rotated daily, I can't imagine them getting huges,
so we'd have to assume that maybe something is writing corrupt logs?
How would I diagnose that?
>
> A couple more points. Is logrotate running at 7:30 simply because it's
> STILL running then?
Not, it seems to be starting around that time.
Runs for about 30 to 60 minutes on any given day.
./tony
--
https://tonybaldwin.info
all tony, all day long...
Attachment:
signature.asc
Description: Digital signature