Date: Mon, 24 Mar 2014 12:06:29 +0600 From: "Eugene M. Zheganin" <emz@norma.perm.ru> To: freebsd-stable <freebsd-stable@freebsd.org> Subject: Re: crash on zpool import - help get data back Message-ID: <532FCB65.6060403@norma.perm.ru> In-Reply-To: <CADbaceL=Gz7bmp6v4dChVPrJnCgLA0SDJyhw3YyQxM_h5dHVGQ@mail.gmail.com> References: <532BEABC.5050808@norma.perm.ru> <532BF531.1050400@norma.perm.ru> <532C0F76.8010605@norma.perm.ru> <CADbaceL=Gz7bmp6v4dChVPrJnCgLA0SDJyhw3YyQxM_h5dHVGQ@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi. On 24.03.2014 08:43, Christian Campbell wrote: > I had a hosed FreeBSD ZFS pool once which I recovered with 'zpool > import -f; zpool export' under an OpenIndiana live CD. > I was able to read all of the data with rsync. I recreated the pool and restored the data read. 'zfs send' was unable to read data - it was crashing while reading 43th gig out of 1400, as well as the scrub was. This whole thing is just sad - clearly something happened with the labels, thus meaning that labels are somehow modified during normal zfs work. Why ? For what reason ? How could the pool become corrupted on a machine with non-corrupted memory ? So my advice to FreeBSD zfs users - scrub periodically and be prepared for some doom to happen. P.S. Saw a couple of minor issues with zfs on Solaris, but never such terrible ones (and it can even swap onto zfs). Eugene.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?532FCB65.6060403>