ZFS panic, system auto reboot

Hi,
I have a system with 4 drives, all IDE, zfs pool (raid z) across the 4, 1 has a small slice carved out with the OS on it. Everything was running stable till a few days ago when I was transferring some files into the pool.

System crashed with a ZFS panic. I thought it was a fluke and it rebooted, but immediately did the same thing. Soon as it tries to make the mount the pool and go active it panics and shuts down. I have tried eliminating one disk and bringing the pool up in a degraded state, no joy.

Booted the system in safe mode from a Live CD, soon as I try to import the pool, panic and crash.

Any ideas?

The crash dump has the following:
Code:
panic: ZFS: I/O failure (read on <unknown> off 0: zio 0x8aab9678 [L0 SPA space map] 1000L/800P DVA[0]=<0:5a0006b000:c00> DVA[1]=
<0:760005e000:c00> DVA[2]=0:910006f000:c00> fletcher4 lzjb LE contiguous birth=30357 fill=1 cksum=826e30d130:8ee58de6107a:5f1171be2ea

I have kmem size and max set to 528. (system is an intel processor) I tried bringing the system up on the cd and then mounting the OS partition and clearing the ZFS cache, no joy.

Thanks for any insight...
TC
 
I would start by importing the pool WITHOUT mounting any underlying filesystems and then try to run a scrub. If this doesn't help, import without mounting any filesystems and then start mounting them one by one, to see which one triggers the panic.
 
Okay, sounds like a logical sequence. So how do I go about importing without mounting?

I have read through my solaris-zfs-administration-guide, gooogled away, and I can't find the switch to import-not mount.

Thanks,
TC
 
I tried a:
zpool import -o mount=none -o failmode=continue -R PrimaryZFS

It returns that everything is online, much like if I had done a zpool import query.

But I can not run a scrub on the pool as zfs still does not see it as being an active pool.

Any suggestions?

Thanks,
TC
 
Back
Top