Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 12 Apr 2017 10:46:49 -0500
From:      CyberLeo Kitsana <cyberleo@cyberleo.net>
To:        "Jason W. Barnes" <jason@barnesos.net>, freebsd-questions@freebsd.org
Subject:   Re: ZFS kernel panic after power outage
Message-ID:  <ea65616a-718a-07f2-f376-579bbef086b2@cyberleo.net>
In-Reply-To: <alpine.BSF.2.11.1704111438030.2744@oom9.barnesos.net>
References:  <alpine.BSF.2.11.1704111438030.2744@oom9.barnesos.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On 04/11/2017 05:38 PM, Jason W. Barnes wrote:
> FreeBSDers:
>
>     After a recent power outage here I have run into trouble whenever I
> try to mount, or export, or list, or do anything at all to one of my ZFS
> file systems (see attached image).  It instantly goes all "panic:
> solaris assert: rs == NULL".  This is my data drive, not my system
> drive, so I commented out "zfs_enable=YES" in my rc.conf and it will now
> boot up after I set the system drive readonly to "off".  But I'd like to
> now try to recover the damaged file system on those drives.
>     The only think that I can think of, and hope for, at this point
> might be to start unplugging one of this three-drive RAID-5 at a time to
> see if that might cleanse the RAID of something that might have been in
> mid-write when the power went out.  Or that could metastasize the
> cancer. Which would be bad.
>     I've seen occasional posts in the past with similar issues, but
> without any firm solutions, so I thought that I'd ask if anyone has any
> insights as to how I might attempt to recover this drive.  Thanks in
> advance if you have any ideas,

ZFS is self-healing - this makes it extremely robust; unfortunately, 
when problems arise that its self-healing cannot correct, it tends to be 
an extremely dire situation indeed.

Having had a look at the code surrounding the above-mentioned assert, it 
doesn't appear to be easily circumventable for recovery. After making a 
copy of the disks involved just in case, you may attempt to import the 
pool in ZFS-on-Linux or Illumos, to avoid possible implementation bugs, 
or import from an earlier, hopefully-consistent txg; however, I suspect 
that you'll have better luck recreating the pool from backups.

-- 
Fuzzy love,
-CyberLeo

<CyberLeo@CyberLeo.Net>
Technical Administrator

CyberLeo.Net Webhosting
http://www.CyberLeo.Net

Element9 Communications
http://www.Element9.net


Furry Peace! - http://www.fur.com/peace/



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?ea65616a-718a-07f2-f376-579bbef086b2>