What happens if master dies

[dignus]

Hey all,

we're thinking about switching our openvz environment to proxmox VE. Can somebody tell me what happens if the master server dies?

We'll replace the box, and then what? Can we make it a master again and re-import all slave cluster nodes?

 

[tom]

Hey all,

we're thinking about switching our openvz environment to proxmox VE. Can somebody tell me what happens if the master server dies?

We'll replace the box, and then what? Can we make it a master again and re-import all slave cluster nodes?

if the master dies, just promote a node to the new master (with pveca).
(you will loose all VM´s stored on the died master, you need to restore them from a vzdump backup).

After you got a new box, just add it to new master. (take a look on the following file for better understanding: /etc/pve/cluster.cfg)

http://pve.proxmox.com/wiki/Proxmox_VE_Cluster

 

[dignus]

Aha, thank you for that. So I'm assuming all cluster nodes are running a webserver and have the "master" data loaded on it?

 

[sengo]

(you will loose all VM´s stored on the died master, you need to restore them from a vzdump backup).

Maybe I just don't get it but what's the point of having a cluster, live migration and all the nice stuff if data is not replicated among the nodes?
Well it's a cluster but NOT a high-availability cluster: if one node dies you loose all the vms and you should restore them manually from backup.

You should keep the storage of the cluster members in sync (DRBD maybe?) or support access to shared storage (ISCS, NFS, FC) and so on.

 

[dignus]

Flexible storage will be implemented in version 2.0, see roadmap on the wiki.

 

[dietmar]

Maybe I just don't get it but what's the point of having a cluster, live migration and all the nice stuff if data is not replicated among the nodes?

Support for shared storage is the next thing we implement.

- Dietmar

 

[sengo]

Support for shared storage is the next thing we implement.

- Dietmar

Keep up the good work then