Ben Donohue wrote:
Hi Slugs,
I have a RH box that once a week the hard disk goes into accessing
continually and won't stop till I literally "pull the plug".
Is this a race condition?
No, that's another problem.
Your problem is "unfairness" -- a process is hammering the
drive and your other processes are not getting a fair share
of disk throughput.
If so then how do I track it down as to why it does it?
Linux is a bit of a nightmare here -- there's no simple utility
showing disk I/Os by process. But a simple starting point is
to ask what is running which usually isn't.
You'll probably find anacron has kicked off the daily tasks
in /etc/cron.daily. The jobs here aren't particularly well
through out for occassionally-used computers and you might
want to shovel some of them into cron.weekly.
(there was no way of gracefully rebooting it).
I've found acpid actually works well here. I have a
/etc/acpi/events/power.conf containing
event=button/power.*
action=/sbin/shutdown -h -f now "Power button pressed"
and it manages to shut down the computer under even high
workloads. I suspect because there's not too much that
needs to be swapped in before the shutdown command is
issued (unlike logging in from a vty).
--
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
