a difficult-to-describe problem involving long write times and enormous load levels

Fajar A. Nugraha list at fajar.net
Sun May 22 00:03:57 EDT 2011


On Sun, May 22, 2011 at 3:55 AM, Daniel Brooks <dlb48x at gmail.com> wrote:
> my only clues are that the
> load average was 180 this morning when I logged in via ssh. The box was
> still operating normally for the root user, as long as I didn't interact
> with a zfs filesystem. The box was even still acting as a gateway for the
> rest of my network. In fact, there was essentially zero cpu usage, and
> although most of my 8 gigs of ram was in use, that wasn't unexpected. There
> was essentially no swap usage at all, which is normal. Is there any way I
> can try to narrow down the actual cause?

I'm guessing that:
- you're disk IOPS bound, or
- you hit some unknown bug in zfs memory management (the last one was
in arc reclaim, but it should be fixed in the version you use)

Can you try:
- "iostat -mx 3" (or some other method that can show disk i/o
utilization) when the load gets big
- limit max arc size using something like this
# cat /etc/modprobe.d/zfs.conf
options zfs zfs_arc_max=134217728

-- 
Fajar



More information about the zfs-discuss mailing list