Rob Mueller wrote:
>
> > > Filesystem is ext3 with one big / partition (that's a mistake
> > > we won't repeat, but too late now). This should be mounted
> > > with data=journal given the kernel command line above, though
> > > it's a bit hard to tell from the dmesg log:
> > >
> >
> > It's possible tht the journal keeps on filling. When that happens,
> > everything has to wait for writeback into the main filesystem.
> > Completion of that writeback frees up journal space and then everything
> > can unblock.
> >
> > Suggest you try data=ordered.
>
> We have a 192M journal, and from the dmesg log it's saying that it's got a 5
> second flush interval, so I can't imagine that the journal is filling, but
> we'll try it and see I guess.
>
> What I don't understand is why the spike is so sudden, and decays so slowly.
> It's Friday night now, so the load is fairly low. I setup a loop to dump
> uptime information every 10 seconds and attached the result below. It's
> running smoothly, then 'bam', it's hit with something big, which then slowly
> decays off.
>
> A few extra things:
> 1. It happens every couple of minutes or so, but not exactly on any time, so
> it's not a cron job or anything
> 2. Viewing 'top', there are no extra processes obviously running when it
> happens
>
If it was this, one would expect it to happen every time you'd written
0.75 * 192 Mbytes to the filesystem. Which seems about right.
Easy enough to test though.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Oct 15 2002 - 22:00:42 EST