GEOM, gvinum, and SATA disks in standby
D X
dragonx at gmail.com
Thu May 18 07:19:59 UTC 2006
After idling my gvinum RAID1 array using ataidle on a pair of WD Caviar
250GB SATA disks, I almost always get a READ_DMA timeout when trying to
access the disk (I'm guessing they take too long to spin up).
At this point one of 3 things will happen (which one it is seems random to
me at this point)
1. Both disks retry fine, and everything's great. Unfortunately, this
happens about 1/3 of the time the disks come out of standby.
2. One of the disks retries fine, but the 2nd disk times out and gets
detached. And I have a degraded array. If I try to restart the downed plex
in gvinum, I get a kernel panic.
3. Both of the disks fail, my OS is partly hung, although I still get
responses to ping packets. My console and terminals are dead though, as is
my httpd. I usually end up rebooting in this case.
I saw a couple of posts online from last year mentioning similar issues. Is
this a known bug that's going to be fixed anytime soon?
More information about the freebsd-geom
mailing list