ZFS - scrub lead to corruption?

[Paul Pathiakis]

From: Graham Todd <gtodd at bellanet.org>

To: Paul Pathiakis <pathiaki2 at yahoo.com>
Cc: freebsd-fs at freebsd.org
Sent: Thu, January 28, 2010 1:24:40 PM
Subject: Re: ZFS - scrub lead to corruption?

Paul Pathiakis wrote:
...
>
> I boot to single-user, make sure that scrubbing is turned off and it
> throws the error within 2 minutes of being up.... I'm SOL unless someone
> (please!!!) can respond and tell me how to roll this back.

I think there was a discussion on -current or this list about using zdb to
force a rollback to a previous uberblock - but I'm not sure if this would
apply to your situation.  Using zdb is a bit more high risk and you might
not be able to rollback that far ...  i.e it might not solve your
corruption issue (the one where /usr/bin/m4 is "there" but "not there"?).

Here's a related solaris discussion you may have seen:

http://www.opensolaris.org/jive/thread.jspa?threadID=85794


****************************************************************************

Graham,

Thanks for the info.  I've realized something that is... disconcerting.  (I had found that article but I don't think the FreeBSD zdb is as robust with features.)

So, I did a quick install of a base OS.  I've backed up everything in /usr that doesn't cause a crash (it's only /usr/bin, /usr/share, and /usr/obj files) onto the new pool.  Many various files in directories and sub-sub-sub-sub... dirs in /usr have files in them that if you touch them *CRASH*

I'm going to zfs destroy the /usr partition and all the subpartitions and recreate it.  I'll restore everything from the new OS /usr.  From there, I'll change my pool mountpoints back to the norm and I will see if I can make the world and install it.

(I'm also going to upgrade the firmware on the drives as well.  I'm going to see if Seagate release a version later than the one I have for the 1500 GB drives)

I'll give the list an update after this.

Paul