Date: Tue, 27 Sep 2011 23:58:49 +0200 From: Johan Hendriks <joh.hendriks@gmail.com> To: peter harrison <four.harrisons@googlemail.com> Cc: freebsd-questions@freebsd.org Subject: Re: GEOM_MIRROR GPT table corrupt or invalid Message-ID: <4E824719.8090102@gmail.com> In-Reply-To: <CAA3eX7YXUM03de8HJbYr2f_VBgvKsy-jSsFXAti=e72i6HK8jg@mail.gmail.com> References: <CAA3eX7YXUM03de8HJbYr2f_VBgvKsy-jSsFXAti=e72i6HK8jg@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
peter harrison schreef: > Hello list, > > I have a problem with a software RAID1. This was created under Freenas 0.7.2 > (using the web interface - FreeBSD 7.3 under the hood) which as I understand > it uses GPT and GEOM_MIRROR. This Freenas is running on AMD64 on an ESXi 4.1 > host with the 2 drives for the mirror passed through as raw disks to the > Freenas guest - if that makes a difference. > > This mirror now fails to mount on boot. Checking the boot messages I see > this straight after the drives are detected: > > GEOM: da1: the secondary GPT table is corrupt or invalid. > GEOM: da1: using the primary only -- recovery suggested > > Further on I see: > > GEOM_MIRROR: Force device data start due to timeout > GEOM_MIRROR: Device mirror/data launched (1/2) > > The mirror is not mounted when booting is completed - no further messages > during boot - mount says: > > freenas:~# mount > /dev/da0s1a on / (ufs, local, soft-updates) > devfs on /dev (devfs, local) > procfs on /proc (procfs, local) > > When I try to mount it manually I get a continually scrolling g_vfs error: > > g_vfs_done():mirror/datap1[READ(offset=-636932692096956416, length =16384)] > error = 5 > > and the guest becomes unresponsive. > > GEOM isn't something I'm really familiar with - so any insight on how this > happened would be interesting. But the main question is, is there anyway of > getting the data off this disk? (I have some backups, but they're not > complete). > > Thanks for any help. > > > > Peter Harrison. > You can not use gmirror and GPT at the same time on the whole disk. Both use the same diskspace to write there meta data. What you can do is to mirror the partitions seperately. A nice howto you can find here http://www.unix-heaven.org/node/20 You can recover the GPT, but then the mirror can not load. if you recreate the mirror after that, the gpt label gets corrupt. regards, Johan
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4E824719.8090102>