Ruslan Sivak
2010-Jul-23 16:10 UTC
[zfs-discuss] ZFS volume turned into a socket - any way to restore data?
I have recently upgraded from NexentaStor 2 to NexentaStor 3 and somehow one of my volumes got corrupted. Its showing up as a socket. Has anyone seen this before? Is there a way to get my data back? It seems like it''s still there, but not recognized as a folder. I ran zpool scrub, but it came back clean. Attached is the output of #zdb data/rt 2.0K sr-xr-xr-x 17 root root 17 Jul 21 17:43 rt # stat rt File: `rt'' Size: 17 Blocks: 4 IO Block: 1536 socket Device: 2d90015h/47775765d Inode: 3 Links: 17 Access: (0555/sr-xr-xr-x) Uid: ( 0/ root) Gid: ( 0/ root) Access: 2010-07-21 13:48:09.485116161 -0400 Modify: 2010-07-21 17:43:13.019153434 -0400 Change: 2010-07-22 15:50:06.222143953 -0400 # cd rt bash: cd: rt: Not a directory # cat rt cat: rt: Operation not supported on transport endpoint a# zfs list NAME USED AVAIL REFER MOUNTPOINT data 992G 828G 47.8K /volumes/data ... data/rt 800G 828G 800G /volumes/data/rt data/rt2 0 828G 800G /volumes/data/rt2 syspool 6.31G 23.0G 35.5K legacy syspool/dump 2.79G 23.0G 2.79G - syspool/rootfs-nmu-000 115M 23.0G 1.32G legacy syspool/rootfs-nmu-001 1.59G 23.0G 1.17G legacy syspool/rootfs-nmu-002 793M 23.0G 1.17G legacy syspool/swap 1.03G 24.0G 16K - # zfs umount data/rt #ls -lahs ... 2.0K drwxr-xr-x 2 root root 2 Jul 22 16:48 rt 2.0K srwxr-xr-x 17 root root 17 Jul 21 17:43 rt2 ... -- This message posted from opensolaris.org -------------- next part -------------- A non-text attachment was scrubbed... Name: zdb.zip Type: application/octet-stream Size: 139160 bytes Desc: not available URL: <http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20100723/14435099/attachment.obj>
Richard Elling
2010-Jul-23 18:14 UTC
[zfs-discuss] ZFS volume turned into a socket - any way to restore data?
On Jul 23, 2010, at 9:10 AM, Ruslan Sivak wrote:> I have recently upgraded from NexentaStor 2 to NexentaStor 3 and somehow one of my volumes got corrupted. Its showing up as a socket. Has anyone seen this before? Is there a way to get my data back? It seems like it''s still there, but not recognized as a folder. I ran zpool scrub, but it came back clean.very strange indeed.> > Attached is the output of #zdb data/rt > > 2.0K sr-xr-xr-x 17 root root 17 Jul 21 17:43 rt > > # stat rt > File: `rt'' > Size: 17 Blocks: 4 IO Block: 1536 socket > Device: 2d90015h/47775765d Inode: 3 Links: 17 > Access: (0555/sr-xr-xr-x) Uid: ( 0/ root) Gid: ( 0/ root) > Access: 2010-07-21 13:48:09.485116161 -0400 > Modify: 2010-07-21 17:43:13.019153434 -0400 > Change: 2010-07-22 15:50:06.222143953 -0400 > > # cd rt > bash: cd: rt: Not a directory > > # cat rt > cat: rt: Operation not supported on transport endpoint > > a# zfs list > NAME USED AVAIL REFER MOUNTPOINT > data 992G 828G 47.8K /volumes/data > ... > data/rt 800G 828G 800G /volumes/data/rt > data/rt2 0 828G 800G /volumes/data/rt2 > syspool 6.31G 23.0G 35.5K legacy > syspool/dump 2.79G 23.0G 2.79G - > syspool/rootfs-nmu-000 115M 23.0G 1.32G legacy > syspool/rootfs-nmu-001 1.59G 23.0G 1.17G legacy > syspool/rootfs-nmu-002 793M 23.0G 1.17G legacy > syspool/swap 1.03G 24.0G 16K - > > # zfs umount data/rt > #ls -lahs > ... > 2.0K drwxr-xr-x 2 root root 2 Jul 22 16:48 rt > 2.0K srwxr-xr-x 17 root root 17 Jul 21 17:43 rt2If at this point data/rt is unmounted, then: mv rt2 rt2-bad zfs mount data/rt -- richard> ... > -- > This message posted from opensolaris.org<zdb.zip>_______________________________________________ > zfs-discuss mailing list > zfs-discuss at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/zfs-discuss-- Richard Elling richard at nexenta.com +1-760-896-4422 Enterprise class storage for everyone www.nexenta.com
Ruslan Sivak
2010-Jul-23 18:34 UTC
[zfs-discuss] ZFS volume turned into a socket - any way to restore data?
On Fri, Jul 23, 2010 at 2:14 PM, Richard Elling <richard at nexenta.com> wrote:> On Jul 23, 2010, at 9:10 AM, Ruslan Sivak wrote: > > > > > > # zfs umount data/rt > > #ls -lahs > > ... > > 2.0K drwxr-xr-x 2 root root 2 Jul 22 16:48 rt > > 2.0K srwxr-xr-x 17 root root 17 Jul 21 17:43 rt2 > > If at this point data/rt is unmounted, then: > mv rt2 rt2-bad > zfs mount data/rt > > > -- richard > >rt2 is actually a copy that i created from a snapshot (unfortunatelly taken after the issue started). I have already tried deleting the folder and mounting again. No luck. The issue is still there. #ls -lahs ... 2.0K sr-xr-xr-x 17 root root 17 Jul 21 17:43 rt ... # zfs umount data/rt #ls -lahs ... 2.0K drwxr-xr-x 2 root root 2 Jul 22 16:48 rt ... # mv rt rt.bad # ls -lahs ... 2.0K drwxr-xr-x 2 root root 2 Jul 22 16:48 rt.bad ... # zfs mount data/rt 2.0K sr-xr-xr-x 17 root root 17 Jul 21 17:43 rt So it''s the volume itself thats shomehow corrupted. I was having some issues with the volume before the upgrade and I remember seeing some issue about duplicate permissions. I have also used chmod and chown to try to fix up some of the issues I was having, and could''ve used them with the wrong options. I dont'' see how it could''ve caused this though. Is it possible the duplicate permissions are causing issues on this volume? Is there a way to clear that? Russ -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20100723/62ffc649/attachment.html>