We have EventstoreDb v24.10 installed in a cluster of three.
I’ve had a read of the documentation around V24.10 for Backups and restores for here but wanted to confirm some things if I can.
We currently have the DB, Index and Logs file on the one data disk (let’s say E:) on a virtual machine in Azure.
At the moment thinking of configuring a nightly full snapshot the E: drive on the first node (1 of 3).
Am I correct in thinking if we needed to restore from this snapshot I can perform the following:
Stop the node (the EventStoreDb service) on that virtual machine
Shutdown the virtual machine
Detach the E: Drive disk
Create a new disk from the snapshot
Attach the new E: drive creating from the snapshot
Start the virtual machine
Start EventStoreDb pointing it to the config file
My questions:
Am I missing anything above?
Can / do we do the same for the other two nodes in the cluster from the same snapshot?
If so, do I stop all 3 nodes at the same time or am I okay to bring them down one by one and restore the E: Drive on each (if needed)?
Is this still something we need to do when taking Azure snapshots nightly?
Any responses or further information you think is important is greatly appreciated.
Personally, I think you can skip this if you are taking regular Azure snapshots, as many of our customers do. But, keep a few things in mind:
If only taking nightly snapshots, your worst-case RPO is approximately 24 hrs. This might not be acceptable, so you might consider more regular snapshotting on the Azure side
You can do the same for all three nodes, which might increase your RTO, but you can also take a snapshot of another node and repurpose it into a new cluster node with (almost) the same configuration (obviously hostnames and things like that which are node-specific would need to be modified after restoring the snapshot)