On Wed, Aug 21, 2002 at 10:02:06AM -0500, Eric Sandeen wrote:
> Russell -
>
> The other thing you might do is compile in kdb, and when it locks up see
> if you can get into kdb. (You'd need to get to a text console, then hit
> the "pause" key).
I tried, but the system was going too slowly to switch. I don't think X
even saw the event, since it didn't switch even after the system
returned to normal.
> On the off chance that this has something to do with your almost-full
> filesystem, you could try unmounting it and seeing if the machine stays
> up - I don't think this is likely to be the problem, though.
Well I did manage to get a ps aux out and it was rather interesting:
root 1885 0.0 0.6 2112 308 ? S 06:25 0:00 /bin/sh -c test
-e /usr/sbin/anacron || run-parts --report /etc/cron.daily
root 1886 0.0 0.5 1248 228 ? S 06:25 0:03 run-parts
--report /etc/cron.daily
root 2179 0.0 0.9 2104 412 ? S 06:30 0:00 /bin/sh
/etc/cron.daily/xfs_fsr
root 2180 0.0 0.9 1316 428 ? S 06:30 0:00 /usr/sbin/xfs_fsr
root 2644 1.8 1.2 1328 544 ? S 07:34 0:01 /usr/sbin/xfs_fsr
root 2646 49.4 2.3 1848 1056 ? R 07:35 0:29 /usr/sbin/xfs_fsr
rhowe 2649 48.0 3.3 3500 1528 ? R 07:35 0:01 ps aux
The same event happening after 15mins of uptime before will probably
have been because xfs_fsr was running as part of cron, took the box
down, so anacron ran the cron job again on startup.
I'm assuming that is the fsr on the almost-full partition which is
locking things up. I will do some more tests. Maybe it isn't always
xfs_fsr that triggers it.. I'm not sure if this always happened at the
same time of day or not, since the clock isn't something I ever really
look at and my sleeping pattern is anything but.
--
Russell Howe | Why be just another cog in the machine,
rhowe@xxxxxxxxxx | when you can be the spanner in the works?
|