kernel: attempt to access beyond end of device

[Erwin123]

This morning a node was down.
Luckily I had a KVM card in it so I rebooted it.
The messages says:

Dec 15 22:06:02 node2 -- MARK --
Dec 15 22:26:02 node2 -- MARK --
Dec 15 22:46:02 node2 -- MARK --
Dec 15 23:06:02 node2 -- MARK --
Dec 15 23:26:02 node2 -- MARK --
Dec 15 23:46:02 node2 -- MARK --
Dec 16 00:06:02 node2 -- MARK --
Dec 16 00:26:02 node2 -- MARK --
Dec 16 00:46:02 node2 -- MARK --
Dec 16 01:00:02 node2 kernel: kjournald starting. Commit interval 5 seconds
Dec 16 01:00:02 node2 kernel: EXT3 FS on dm-5, internal journal
Dec 16 01:00:02 node2 kernel: EXT3-fs: dm-5: 6 orphan inodes deleted
Dec 16 01:00:03 node2 kernel: EXT3-fs: recovery complete.
Dec 16 01:00:03 node2 kernel: EXT3-fs: mounted filesystem with ordered data mode.
Dec 16 01:09:48 node2 kernel: attempt to access beyond end of device
Dec 16 01:09:48 node2 kernel: dm-5: rw=32, want=10640669328, limit=1951596544
Dec 16 01:09:54 node2 kernel: attempt to access beyond end of device

etc....(these lines go on forever)

From google I gather its probably a problem with the filesystem?
Although I had a problem with slow performance this is the first real problem I have with it. Could this be the raid card or also something more software related? Any other suggestions?
Thanks in advance.

 

[dietmar]

Maybe that error is from a snapshot device with space overflow. I would suggest to stop macking backups using lvm snapshots until you find out why your disk are that slow.

 

[Erwin123]

Okay thanks.
I adjusted yesterday the vzdump size: to 2048 from 3072 as an experiment. I tried to set it first at 4096 but the system refused that.

 

[Mr.Embedded]

I have seen this before with a Raid array no Proxmox related. The partition table got corrupted on reboot and shrink the partition. The cause was using an older version of parted that had a bug with debian, when creating the large GPT partition I was using. I updated to latest and greatest parted and remade the partition table without reformating and all better. Not sure if this will resolve your issue but I figured I would mention it.