Periodic update of a replications server

[yunmkfed]

Hi,

My setup is a non-critical lab environment. Having said that I have VMs I treasure and don't want to loose (or redo).

With my VMWare I periodically stop a VM, RAR a volume and archive it. If I screw up the running VM, I transfer the archive, expand it and use it. I understand that this approach will result in data loss but I accept the risk and it does not bother me.

On Proxmox I have setup backups. My first question is how can I test this out without destroying the original. With VMWare I could extract to a different directory and when I attempt to run the VM I indicate that the machine has been copied rather than moved (generating a new MAC).

I do not have HA needs, but read about replication. My understanding of replication is that one has a number (2 in my case) PVE's and the VM is copied from one cluster to the other. With replication, if the main proxmox server goes down, would I be able to start the replicated machine and continue from there?

Thanks

 

[jsterr]

Yes. Thats how replication works. If server 1 goes down, vm starts on server 2 with the data from the last sync. Automatically if you use ha, non automatically if you move the qm-file /etc/pve/qemu-server manually to the right directory. (this is usually done by the ha-manager) needs to be done manually if you dont use ha. For example, if node 5 is down, you need to move the file from here with a mv-command. This will result in having the vm-config present on the new node.

root@PMX4:/etc/pve# cd nodes/PMX5/qemu-server/

root@PMX4:/etc/pve/nodes/PMX5/qemu-server# ls
102.conf  103.conf  104.conf  105.conf  107.conf  108.conf  110.conf  117.conf  122.conf  124.conf  125.conf  300.conf  9002.conf

to here:

root@PMX4:/etc/pve/nodes/PMX5/qemu-server# cd /etc/pve/qemu-server

root@PMX4:/etc/pve/qemu-server# ls
106.conf  111.conf  120.conf  121.conf  123.conf  150.conf  151.conf  152.conf  153.conf  154.conf

Regarding backups you can restore to a different vm-id and youll get new mac-address by setting the "unique" chechbox.