I/O corruption with vzdump?

[resignation]

Good day,

I am having an apparent disk corruption issue when using vzdump. I'm not certain if it started after upgrading to Proxmox VE 1.2, but during an automated backup that was set with the webGUI, I get kernel messages such as this for each VM (KVM or OpenVZ) that is backed up.

kjournald starting.  Commit interval 5 seconds
EXT3 FS on dm-3, internal journal
ext3_orphan_cleanup: deleting unreferenced inode 24298041
ext3_orphan_cleanup: deleting unreferenced inode 24298040
ext3_orphan_cleanup: deleting unreferenced inode 24298039
ext3_orphan_cleanup: deleting unreferenced inode 24298038
ext3_orphan_cleanup: deleting unreferenced inode 24297938
ext3_orphan_cleanup: deleting unreferenced inode 24068240
EXT3-fs: dm-3: 6 orphan inodes deleted
EXT3-fs: recovery complete.
EXT3-fs: mounted filesystem with ordered data mode.

During the backup run, I can see a new device being mounted and accessed..

/dev/mapper/pve-vzsnap
                       1.9T    14G   1.9T   1% /mnt/vzsnap

Is this normal behavior? Otherwise all looks fine and the backups seem to restore properly, but whenever I see errors like that I start to worry. My hardware is as follows..

Gigabyte GA-EP45-DS3R (P45/ICH10R)
Intel Core 2 Quad Q9400
3Ware 9650SE w/ 4x1TB SATA RAID10
Realtek RTL8111 (2) and Intel 82541PI (2) Nics

I have no problems restoring the backups, all works as expected. Just nervous about those errors, has anyone seen this before? I am hesitant to rebuild the box from scratch just to see if it fixes the problem.

Thanks!

 

[dietmar]

Is this normal behavior?

Yes, its normal (we make a snapshot of the filsystem, and the snapshot needs to cleanup the journal befor it is mounted).

- Dietmar

 

[resignation]

So there is no need to monitor this filesystem for errors or run fsck's on it? Also I assume the more data you have to backup via vzdump, the more errors the filesystem will spit out?

Thanks for the quick response!

 

[dietmar]

So there is no need to monitor this filesystem for errors or run fsck's on it?

As mentioned above, it is a lvm snapshot of another lvm volumen.

Also I assume the more data you have to backup via vzdump, the more errors the filesystem will spit out?

No, not really.

- Dietmar

 

[mot]

dietmar,

Can you give a little bit more information?

We have the same issue when we make vzdump using snapshots, and, for example, if we backup 3 VM's the logwatch shows us:
....
3 Time(s): ext3_orphan_cleanup: deleting unreferenced inode 4980741
3 Time(s): ext3_orphan_cleanup: deleting unreferenced inode 4981451
3 Time(s): ext3_orphan_cleanup: deleting unreferenced inode 4981482
....
As far as I know this happens when the inode point to an incorrect sector of disk.

You are 100% sure there is no need to do any periodic checks?

 

[dietmar]

Again, I think those messages are from the snapshot. If you weant to be sure do a reboot - then the original filesystem is checked.