Corey

>> -----Original Message-----
>> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
>> Sent: Friday, September 12, 2008 4:34 PM
>> To: Leopold, Corey
>> Cc: zfs-discuss@opensolaris.org; [EMAIL PROTECTED]
>> Subject: Re: [zfs-discuss] ZPOOL Import Problem
>>
>> Corey,
>>
>>> I ran into an odd problem importing a zpool while testing avs.  I  
>>> was
>>> trying to simulate a drive failure, break SNDR replication, and then
>>> import the pool on the secondary.  To simulate the drive failure is
>>> just offlined one of the disks in the RAIDZ set.
>>>
>>
>> Are all constituent volumes of a single ZFS storage pool in the same
>> SNDR I/O consistency group?
>>
>
> Yes they were.
>
> One thing to note that shows some of my unfamiliarity with SNDR is  
> that
> I actually deleted the replication set config before trying to mount  
> on
> the secondary "sndradm -g group -nd"
>
> Since then I have been throwing them into logging mode "sndradm -g  
> group
> -nl" and haven't had a problem in similar tests. (i.e. offlining a  
> drive
> before making the secondary active).
>
> I don't believe that deleting the SNDR replication configuration  
> should
> have made the ZPOOL invalid though.  So there may still be a bug
> somewhere.

If at the time the SNDR replica is deleted the set was actively  
replicating, along with ZFS actively writing to the ZFS storage pool,  
I/O consistency will be lost, leaving ZFS storage pool in an  
indeterministic state on the remote node.  To address this issue,  
prior to deleting the replicas, the replica should be placed into  
logging mode first. Then ZFS will be left in I/O consistent after the  
disable is done.

> Corey

Jim Dunham
Engineering Manager
Storage Platform Software Group
Sun Microsystems, Inc.

_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss

Reply via email to