Ubuntu 9.10Server - rename(/dev//.udev-tmp, /dev//) failed. invalid cross-device link

[Guest Guest]

Got abit of a problem.

 

Our server room was red hot and all the virtual servers were going potty. I decided a reboot was in order so i proceeded to reboot the XenServers and for good measure give the San a reboot. What a mistake that was!

 

The San is an Ubuntu Server 9.10 running LVM, with the Xen SRs presented over NFS.

 

The San now wont boot back up, giving the error;

 

udev-work[6310]: rename(/dev//.udev-tmp,/dev//) failed. invalid cross-device link

 

I tried booting from CD to do a recover; it hung at the above error. Fair enough i thought, ill just have to rebuild as it needs to be up ASAP and rather than mess around...

 

Set it away to reinstall and it gets stuck at "Setting up partitioner" with the same error message in teh debug console. Ok, if i wipe the OS partition in the RAID controller config and start again... Again, same error.

Obviously its reading from teh data partitions to check what LVs are there but seems to get stuck. grr!

 

 

Any ideas on what to do to fix this? I have backups of all the VMs but id rather not have to reimport everything as that takes ages!

 

Help please

Edited May 18, 2010 by Guest

[Guest Guest]

Bump..

[webman]

You're really running a mission-critical server on a non-LTS version of Ubuntu?

 

I'd advise a fresh install of 10.04 and see if you still have those errors at setup. If you do, can you disconnect the LVM volumes that are used as Xen SRs while installing Ubuntu, then re-connect them afterwards and configure their mountpoints?

[Guest Guest]

You're really running a mission-critical server on a non-LTS version of Ubuntu?

 

I'd advise a fresh install of 10.04 and see if you still have those errors at setup. If you do, can you disconnect the LVM volumes that are used as Xen SRs while installing Ubuntu, then re-connect them afterwards and configure their mountpoints?

 

Well, its a long story, but the jist of it is;

 

We had a major problem with our Sans so i was forced to rebuild as quickly as i could using whatever was availible to hand, with a new Sun San to be ordered the following week. That didnt happen and we were told to order new physical servers instead as "£9000 is too much to be spending on a San, but £9000 on 3 servers is ok"

That all started back in february, its now May and we have only just ordered yesteraday!!!

 

Obviously i havnt been able to take the San down to put a new OS with failover to th 2nd san etc etc as it would disrupt lessons...

 

 

Anyway the problem is sorted (sortof). Its all back up and running. The solution; leave it overnight to boot. It now boots with errors, but eventually does get past anfd loads up and works fine. Installed 10.04 but taht made no difference to the actually cause of the problem. Eitherway, the servers are on the way... fingers crossed it holds out till then!

Edited May 19, 2010 by Guest

[webman]

I feel your pain

 

At the bottom of a Ubuntu Launchpad bug report (Google the error description) there's a script modification you might be able to make to turn the errors off; but if things are in hand to move and it's sort-of working now, I would just leave it. Don't look/touch/breathe near the server just in case