Question about eventual node failure during ZFS replication in 2-node cluster

[fmoreira86]

Hi everyone,


I have a Proxmox cluster with 2 nodes, both using local ZFS storage. Some VMs are configured for ZFS replication between the two nodes.


My question is:
If the node running a given VM (let's call it VM X) fails during a ZFS replication process, will the second node have a valid and usable copy of the VM?
Or could the replica be inconsistent or corrupted because the replication was not completed at the time of the failure?


Thanks in advance for your help!

 

[guruevi]

ZFS replication is atomic. So until it completes, you just have the previous version of the replica.

 

[fmoreira86]

ZFS replication is atomic. So until it completes, you just have the previous version of the replica.

Thank you for your clarification!

 

[leesteken]

I have a Proxmox cluster with 2 nodes, ...

Please be aware that when one node fails or becomes unreachable, the other will go read-only (and reboot in case of HA?) and can't start VMs as it does not have quorum. You really need a third vote, like a QDevice for example. Maybe you already know all this but otherwise it's all in the Proxmox manual.

 

[fmoreira86]

Please be aware that when one node fails or becomes unreachable, the other will go read-only (and reboot in case of HA?) and can't start VMs as it does not have quorum. You really need a third vote, like a QDevice for example. Maybe you already know all this but otherwise it's all in the Proxmox manual.

I've a qdevice.

Cheers.