Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 14 Feb 2017 22:05:56 -0800
From:      Chris Stankevitz <chris@stankevitz.com>
To:        FreeBSD Questions <freebsd-questions@freebsd.org>
Subject:   Re: Stale zpool from a prior life in "zpool import"
Message-ID:  <ed943f5b-6709-3965-0cdd-24d0153a18b2@stankevitz.com>
In-Reply-To: <3aff436c-4e15-ec41-5b54-532cc036fdc0@toyon.com>
References:  <3aff436c-4e15-ec41-5b54-532cc036fdc0@toyon.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On 2/14/17 12:45 PM, Chris Stankevitz wrote:
> How do I remove the UNAVAIL zpool?

Here's how I solved this:

1. use "zdb -l /device" on my devices to figure out which ones held 
references to these ghost zpools.  I had to call this on my raw disks, 
the partitions, and the decrypted .eli devices to catch them all

2. for each of the devices discovered with ghost zpool data, I ran 
"zpool labelclear -f /device".  Unfortunately this destroyed the geli 
metadata.  grrr.

3. I was careful about the order in which I called labelclear so that if 
devices were destroyed (which they were) a resilver would have brought 
my pool back.  However, the geli folks backed up the metadata to 
/var/backups during init.  I used "geli restore /var/backups/foo.eli 
/dev/gpt/foo" on the affected partitions to restore the encrypted 
partitions.  I wonder what else "zpool labelclear" could have destroyed 
(gpt???).

Chris



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?ed943f5b-6709-3965-0cdd-24d0153a18b2>