recover scenario with external storage cluster

[kablubr]

Hi guys,

I am about to build an external storage cluster (2 servers) with HAST + CARP. The kind Active/Passive. They will be connected to a 3 nodes cluster of proxmox with HA.

My concern is, if a storage node fails, the other will assume with some seconds of delay with the same IP, however probably all the vms will break/crash.

Did proxmox will restart all those VMs?

Is there some way of proxmox recover from this process? HOW?

Thank you very much!!

 

[wahmed]

What are you using to build Storage Cluster? DRBD/CEPH/FreeNAS/GlusterFS?

My understanding is Proxmox will not start any VM if the cause is storage failure. Since External Storage Cluster really is not controlled or maintained by Proxmox itself. It will simply lost connection with the storage and produce errors within the cluster. You will manually have to restart all VMs.

I believe storage like FreeNAS, Nexenta are the ones where you will need active+passive. You can achieve the same goal with less headache with storage like GlusterFS/CEPH and may be DRBD. Gluster, CEPH gives you real time redundancy where a complete node failure does not bring down everything. Your VMs will simply keep working like nothing happened.