Maurilio Longo wrote:
Hi,
I have a pc with a MARVELL AOC-SAT2-MV8 controller and a pool made up of a
six disks in a raid-z pool with a hot spare.
...
Now, the problem is that issuing an
iostat -Cmnx 10
or any other time intervall, I've seen, sometimes, a complete stall of disk
I/O due to a disk in the pool (not always the same) being 100% busy.
...
In this case it was c2t2d0 and it blocked the pool for 30 or 40 seconds.
/var/adm/messages does not contain anything related to the pool.
What can it be?
This usually means you have either a driver bug, a bad controller, or a bad disk
The marvell driver bug sometimes manifested in this way, but you would have seen
bus resets in your error logs.
Given you have exactly one outstanding transaction on the "stuck" disk, I
suspect the disk is busy doing error recovery.
Speaking from my recent extremely painful experience, replace that disk ASAP.
--
Carson
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
