help - reinstallation problems - udevd: timeout: killing in console

[pixelplumber]

Hi, this is a continuation of the issues I've had in the thread "upgrade interrupted, dpkg half-configured server not booting".

My problems have carried over into the reinstallation of proxmox. I'm trying to use the latest 3.1 ISO to reinstall to /dev/sda.

During booting of the installer CD I receive the udevd timeout killing errors repeatedly (which I also would receive on the corrupted install). It does eventually boot into the installer however it fails at the partitioning stage, perhaps due to the previous errors?



Now, given I had the same errors in my corrupted install after an upgrade (enterprise repository), I'm wondering if its a kernel issue and compatibility with my hardware?

We have a Dell T620 server with PERC 710 RAID card. This machine has been running 2.3 and 3.0 for the last year without problems until the most recent upgrades.

Are there any boot options I can try? I'm at a bit of a loss now. Going to try and reinstall with 2.3.

 

[pixelplumber]

Second install attempt with 3.1 ISO. It got further this time, past the disk setup and all the way through the install process until the NFS-common package and then appears to hang indefinitely.

Given this is booting off a CD and not the corrupted install on disk and getting similar problems, there must now be some sort of kernel hardware issue with this dell 620 and the latest proxmox correct?



Going to try 2.3 now.

 

[pixelplumber]

The saga continues. Booting from the downloaded 2.3 ISO locks up again. Thinking perhaps it is hardware problem now, but unsure of how to find what it is causing it. Had been running fine for the last year.



My next step to try is installing without using the iDrac remote console, and only use local console in case that has something to do with it.

And failing that to try a vanilla centos install to see if I can confirm similar errors accross installed distros with different kernels.

 

[mir]

My best guess is disk or RAM problems. What does smartmon have to say about your disks? Does running a memory tester display any memory problems?

 

[pixelplumber]

In the interests of not leaving the thread hanging the problems seem to have been solved. A drive was predicted to fail (had not actually failed yet) but I suspect either it was somehow exposing a RAID firmware bug, or filling up the controllers buffer with IO timeouts, as once the disk was replaced and the firmware upgraded, proxmox 3.1 could be reinstalled and setup, with no more randome IO lockups in the kernel (even when CD booting).

Fortunately by running my VZ and KVM images on a custom LVM mount the reinstall didn't blow away the data and after manually re-tagging the logical volumes and hooking them back up to recreated virtual machine configs I had the old data back and running.