zfs discuss - Sep 2008 - resilver being killed by ''zpool status'' when root

is there a bug for the behavior noted in the subject line of this post?

running ''zpool status'' or ''zpool status -xv''
during a resilver as a non-privileged user has no adverse effect, but if i do
the same as root, the resilver restarts.

while i''m not running opensolaris here, i feel this is a good forum to
post this question to.

(my system: SunOS filer1 5.10 Generic_137112-07 i86pc i386 i86pc)

thanks,
blake
--
This message posted from opensolaris.org

>>>>> "bi" == Blake Irvin <blake.irvin at
gmail.com> writes:
    bi> running ''zpool status'' or ''zpool status
-xv''
    bi> during a resilver as a non-privileged user has no adverse
    bi> effect, but if i do the same as root, the resilver restarts.

I have this in my ZFS bug notes:

From: Thomas Bleek <bl at gfz-potsdam.de>
 "zpool status" is resetting the resilver process of a spare drive.
 Only the spare drive resilver is disturbed, an "normal" resilver not!
 Because I have a script running every 10 minutes to check some
 things (also zpool status) the resilver did never complete:-(

was yours a resilver onto a spare drive, or a manually-initiated
resilver?
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 304 bytes
Desc: not available
URL:
<http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20080923/7da9d96c/attachment.bin>

I was doing a manual resilver, not with spares.  I suspect still the issue comes
from your script running as root, which is common for reporting scripts.
--
This message posted from opensolaris.org

I''ve confirmed the problem with automatic resilvers as well.  I will
see about submitting a bug.
--
This message posted from opensolaris.org

Looks like there is a closed bug for this:

http://bugs.opensolaris.org/view_bug.do?bug_id=6655927

It''s been closed as ''not reproducible'', but I can
reproduce consistently on Sol 10 5/08.  How can I re-open this bug?

I''m using a pair of Supermicro AOC-SAT2-MV8 on a fully patched install
of Solaris 10 5/08, with a 9-disk raidz2 pool.

The motherboard is a Supermicro H8DM8-2.
--
This message posted from opensolaris.org

Blake Irvin wrote:
> Looks like there is a closed bug for this:
> 
> http://bugs.opensolaris.org/view_bug.do?bug_id=6655927
> 
> It''s been closed as ''not reproducible'', but I
can reproduce consistently on Sol 10 5/08.  How can I re-open this bug?
Have you tried to reproduce it with Nevada build 94 or later? Bug 
6655927 is closed as not reproducible because that part of the code was 
rewritten as part of fixing 6343667, and problem described in 6655927 
was not reproducible any longer.

If you can reproduce it with build 94 or later, then the bug 6655927 
probably worth revisiting.

victor

Pls pardon the off-topic question, but is there a Solaris backport of the fix?


On Tue, Oct 21, 2008 at 2:15 PM, Victor Latushkin
<Victor.Latushkin at sun.com> wrote:
> Blake Irvin wrote:
>> Looks like there is a closed bug for this:
>>
>> http://bugs.opensolaris.org/view_bug.do?bug_id=6655927
>>
>> It''s been closed as ''not reproducible'', but
I can reproduce consistently on Sol 10 5/08.  How can I re-open this bug?
>
> Have you tried to reproduce it with Nevada build 94 or later? Bug
> 6655927 is closed as not reproducible because that part of the code was
> rewritten as part of fixing 6343667, and problem described in 6655927
> was not reproducible any longer.
>
> If you can reproduce it with build 94 or later, then the bug 6655927
> probably worth revisiting.
>
> victor
> _______________________________________________
> zfs-discuss mailing list
> zfs-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
>

As jritorto is noting, I think the issue here is whether the fix has been
backported to Solaris 10 5/08 or 10/08.  It''s a nasty problem to run
into on a production machine.  In my case, I''m restoring from tape
because my pool went corrupt waiting for resilvers to finish which were getting
killed by a crontabbed reporting script.  Looks like a perfect time for a
backport.
--
This message posted from opensolaris.org