starting dlm_controld stuck

[TrevorJ]

Dear members,

I have just clicked the upgrade button in Proxmox and my server stopped responding.

After a reboot, the server get's as far as Starting dlm_controld .... INFO task ksmtuned:3792 blocked for more than 120 seconds.

Please see attached screen capture.

Does anyone know how to fix this issue, please help it's a production server.

Thanks.

 

[dietmar]

Seems you have a complex setup - cluster, DRBD, ..? Hard to help without any information.

 

[TrevorJ]

Hi,


I initially set-up a 2 node cluster with the view of making HA work later on.



I wanted to make sure that the VM ids did not overlap from the start and therefore created the cluster.



About 4 months ago, I set-up a DRDB partition on my two nodes using the following tutorial and have been testing performance since.



http://pve.proxmox.com/wiki/Two-Node_High_Availability_Cluster



I used a third server with a Quorum Disk to create the DRDB config.



I never got around to setting up Proxmox HA with Fencing.



As previously mentioned, I have had no problems with either of the nodes and have been rebooted numerous times.



The issue with the dlm_controld only happened once I upgraded last night.



What does this process actually do? Is it possible that iscsi is now broken on the node and it can't connect to the Quorum partition on the third server?



Can someone please explain what the error is and what it means?



Many Thanks,

 

[dannyman]

Hello,

Did you ever solve this problem? I have several clustered nodes that have a rough time starting up, often hanging at dlm_controld, as described here. Sometimes I can get them to boot cleanly if I remove he network cable. Tonight, I can boot the problem node without network then sometime after I restore the network the node locks up again. :<

-danny