Failed reboot - VM stuck frozen (HA)

[Blaz Repas]

Hello!

I have a 7 node cluster and a software/lrm HA. I have just done maintenance on one of the nodes and have rebooted it. Due to some hardware failure it failed to boot. The VMs that were on that node are in a frozen (HA) state and I can not migrate or start them.

Is there a way to convince the HA to consider the rebooted node as failed and to migrate the VMs to running nodes - basically to unfreeze the frozen VMs?

I am running:
Kernel Version
Linux 4.13.13-4-pve #1 SMP PVE 4.13.13-35 (Mon, 8 Jan 2018 10:26:58 +0100)

PVE Manager Version
pve-manager/5.1-42/724a6cb3

Any help is appreciated...

 

[dietmar]

see https://pve.proxmox.com/wiki/High_Availability#ha_manager_error_recovery

 

[Blaz Repas]

Thank you for the reply.
However, the VM is not in an error state, but rather in a frozen state. I have already tried putting it into disabled, started, stopped states but it still remains in the frozen state.

 

[t.lamprecht]

However, the VM is not in an error state, but rather in a frozen state. I have already tried putting it into disabled, started, stopped states but it still remains in the frozen state.

Can you try to remove and then re-add it to HA again?

 

[tonci]

Hello , I'm facing the same issue ... I have 2-node HA cluster and Prox-backup server as 3rd node with qnetd service installed. The cluster is quorate (3 votes) and everything seemed to work fine (prox 7.1.5). During playing with HA features cluster somehow got into frozen state ... both nodes idle and CT got in frozen state. CT state could not be changed like above and remove-attempt (like suggested) got stuck in "deleting" state forever. Restarting both nodes did not changed "deleting" state and next try removing the same CT resulted with error-mess "cannot delete service : not HA managed! BUT, the service(CT) has been actually freed/unlocked and fro that point on I could manipulate it (start/stop/migrate/...) ... BUT I cannot put it in HA anymore .. not listed ....
Thank you in advance for you support ...

best regards
Tonci

 

[tonci]

This is the point I got stuck ... CT:101 is available out of the HA but cannot be put back.

If I want to put another VM in HA it's state becmoes "queing" forever ...