Migrate CT from dead cluster node

[iJanki]

Hello all,

I'm new to proxmox. Currently I've a test setup with two nodes in cluster (no HA).
I have a openVZ CT on a shared external NFS. I can migrate the CT successfully from one node to another.
But If a take down the node with the active CT, I can't migrate anymore to the other node.
Is it supposed to work like this?

Thanks.

 

[dietmar]

But If a take down the node with the active CT, I can't migrate anymore to the other node.
Is it supposed to work like this?

Yes

 

[iJanki]

So to achieve some sort of redundancy the only way is to use HA? In proxmox 1.x we used to create the configuration of a VM on both nodes, keeping one turned off, and turning it on in case the other node failed.
In case HA when a node goes down, the CTs inside are moved automatically to other nodes?

 

[tom]

1.x series has no HA, all configs and also containers are just stored on local disks.

2.x got HA, config files are stored in http://pve.proxmox.com/wiki/Proxmox_Cluster_file_system_(pmxcfs)

 

[iJanki]

I know 1.x has no HA. But I could achieve some sort of redundancy using a shared redundant storage and creating the same VM configuration on both nodes. When one cluster node goes down I can start the VM on the other node manually.

Can I do the same with Proxmox 2 without HA? (I.E. manually recover the VM on the dead node and starting it on the other node)

 

[dietmar]

Can I do the same with Proxmox 2 without HA? (I.E. manually recover the VM on the dead node and starting it on the other node)

You can manually move the configurating file:

# mv /etc/pve/nodes/<node1>/qemu-server/<VMID> /etc/pve/nodes/<newnode>/qemu-server

 

[iJanki]

Thanks!

The mv gives

mv: cannot move `proxmox2a/openvz/100.conf' to `proxmox2b/openvz/100.conf': Permission denied

How can i write on /etc/pve? I'm root.

 

[tom]

looks like you do not have quorum in you cluster, therefore the configuration filesyste is readonly to protect edits.

if you are sure what you do, set the expected votes to 1, so that a standalone machine have quorum with just one vote.

pvecm e 1

 

[kankamuso]

looks like you do not have quorum in you cluster, therefore the configuration filesyste is readonly to protect edits.

if you are sure what you do, set the expected votes to 1, so that a standalone machine have quorum with just one vote.

pvecm e 1

This command complains about that cman is not running... because there is no quorum... any more choices?

 

[net-4-it]

This command complains about that cman is not running... because there is no quorum... any more choices?

I had the same problem. First time I reinstalled the system for fixing the problem. I had a failure in the /etc/pve/cluster.conf. If there is a failure, cman can't start and the quorum can't be connected, so it makes the files in /etc/pve/ read-only.
Now I found out, that there is an other cluster.conf... Try the following:
1. service pve-cluster stop
2. nano /etc/cluster/cluster.conf --> change what ever you have to change and don't forget to count up the config_version="N+1" if you don't count it up, the file will be overwritten by the /etc/pve/cluster.conf
3. (Sorry, I was to long in the windowsworld, so): reboot the system
4. If the system is up again: service cman status
5. If cman is online and you see, that you have write access in /etc/pve/, cman has been loaded with the config in /etc/cluster/cluster.conf
6. now you have to make the changes in the /etc/pve/cluster.conf
7. reboot again