I have a zpool that has been plagued due to physical disk failures.
The zpool consists of two raidz2s. There are a few disks that have
been removed from the zpool due to failure, and otherwise, the data on
failing disks has been copied to new media and the few (<10 per disk)
blocks that were unreadable and filled with zeros.
I have reached a configuration that, while short of the full number of
disks, has no failing disks involved. I have been able to import the
zpool and access all data without errors. The problem is that in the
process of resilvering, the system panicked. Now I have a pool I can't
import, not even with the following in /etc/system:
set aok=1
set zfs:zfs_recover=1
And yes, I am remembering to delete /etc/zfs/zpool.cache. And I tried
importing read-only. And I tried a live disc.
I know the data is intact (there's enough parity) and every disk is
functioning properly. But some kind of subtle corruption is preventing
it from importing and resilvering at this point.
If I return some of the failing media to the zpool, I am able to
import it. However, the failing media results in either glacially slow
access times and/or the disk is faulted at some point. This basically
doesn't help me.
Is there any way to overcome this crashing problem?
My only other option is to copy the failing disks to new media, block
level, filling any unreadable blocks with zeros. But this is time
consuming and frankly, I am running out of disk media. And the whole
point of a zpool is so that it manages the storage so I don't have to
do stupid things like this.
Ideas anybody?
-Galen
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
