Proxmox VM DR Scenario

[emregozturk]

Hello everyone,

I have two different Proxmox VE clusters in my data center located in two different geographies. I want to replicate for DR of VMs running on Proxmox. Currently, I can provide this with file system replication on Storage, but it is not very useful. Do anyone protect against DR condition with Proxmox? With which product and how can I replicate virtual servers between two sites as I explained above?

 

[aaron]

As usual, it depends

Do you use the Proxmox Backup Server? Then it would be quite easy, have a PBS in each location and set up a sync in between them (namespaces are useful to separate locations). Then you can (live) restore from the latest backup.

Do you have a Ceph cluster, then RBD mirroring might be a way, but needs you to copy the VM config files manually and also handle a failover more manually.

This is something that will hopefully become easier in the future.

Currently, I can provide this with file system replication on Storage, but it is not very useful.

If you sync the storage in both sites and configure it as storage to place the VMs on (network share?) you could copy the VM configs to the other site on a regular basis. The replication/sync of the storage needs to be stopped/reversed in a DR situation.

 

[emregozturk]

As usual, it depends

Do you use the Proxmox Backup Server? Then it would be quite easy, have a PBS in each location and set up a sync in between them (namespaces are useful to separate locations). Then you can (live) restore from the latest backup.

I'm using PBS yes but as you can appreciate backup is not a DR solution because RPO and RTO times will be too high. But I still want to learn and try Is it possible to share a document for this?

Do you have a Ceph cluster, then RBD mirroring might be a way, but needs you to copy the VM config files manually and also handle a failover more manually.

I don't have

This is something that will hopefully become easier in the future.

Waiting Impatiently.

If you sync the storage in both sites and configure it as storage to place the VMs on (network share?) you could copy the VM configs to the other site on a regular basis. The replication/sync of the storage needs to be stopped/reversed in a DR situation.

I'm doing this partially, the missing part is just how can I copy the VM configuration? I do it manually by creating a VM on the DR site then attaching a disk via replicated nfs share. I also use L2 VPN or VxLan for Network, there is no problem here.

 

[aaron]

The VM config files are located in /etc/pve/qemu-server, or more precisely in /etc/pve/nodes/<node>/qemu-server. /etc/pve/qemu-server is a symlink to /etc/pve/nodes/<current node>/qemu-server.

For containers, it is similar, but instead of the qemu-server you have the lxc directory.
My copying the config files to the other cluster, they should be able to start if the storage and network configuration is the same.

Making sure that only 1 instance is running and that the replicaton of the disks is only working in one direction is currently up to you to guarantee

 

[emregozturk]

Thank you for detailed info.

Is there an application that can be a DR scenario with a 3rd party tool other than the information you have provided? Veeam, RP4VM and Zerto can be used for VMware guest VMs. Actually, I'm wondering if there are any tools similar to these.

 

[aaron]

I cannot answer your question about 3rd party options, but I think one thing that you might have missed, is the option of a live restore for VMs when restoring from a PBS server. This will start the VM a few moments after the restore has been started while the restore is ongoing.

This could be an option to get the infrastructure up and running quite fast.

Ideally, the PBS has fast SSD storage and a fast connection to the PVE server, as not-yet-restored data is redirected to the PBS server.

 

[emregozturk]

I cannot answer your question about 3rd party options, but I think one thing that you might have missed, is the option of a live restore for VMs when restoring from a PBS server. This will start the VM a few moments after the restore has been started while the restore is ongoing.

This could be an option to get the infrastructure up and running quite fast.

Ideally, the PBS has fast SSD storage and a fast connection to the PVE server, as not-yet-restored data is redirected to the PBS server.

Understood thanks. Can you provide me a document or article link so that I can configure the PBS?

 

[aaron]

https://pve.proxmox.com/pve-docs/pve-admin-guide.html#_live_restore

If you restore a VM on Proxmox VE from a backup that is on a Proxmox Backup Server, you have the additional "Live Restore" checkbox in the web UI, or can use the option on the CLI as explained in the docs.

 

[renegade247]

I am very new to the use of Proxmox. I am trying to set up a DR site.
I have a few VM's but its largely a K8s environment in my current production environment. What are my options in setting up a DR site?