I'm really starting to get into SRM now 🙂
Is there a way to have a recovery plan do a snapshot of a VM before it powers it on in a recovery plan?
Here's the scenario. We have an app with a primary and secondary node. They fail back and forth all the time. For recovery, the primary needs to be brought up first otherwise, there could be data loss. There's no way to tell which is primary until you power a node on. If you power the wrong one first, there's data loss. I want SRM to do a snapshot before powering them on. This way, if we go into the app and the wrong node was powered on first, we can just revert to snapshot and power the other node on first. I know we can just cleanup and run the recovery plan again but, there are a couple of other apps being recovered with the same plan.
In the recovery plan you can configure Priorities for VMs as well as dependencies so if you know what is the VM that should start first you should be able to configure that in the recovery plan so the second VM waits until the first one is fully up and running
I understand that and I have recovery plan setup with the primary data node powering on first. The problem is, the node that is primary can switch and I would never know it until I try to recover them at the DR site. Node 1 is primary right now and everything should work fine but, if node 2 becomes the primary down the road and we run the recovery plan, it's going to power on the wrong node first.
I guess the other option would be to somehow have the app send an alert that triggers a script to edit the recovery plan each time the nodes fail over but, that seems like a lot more work than just creating a snapshot in the unlikely event we ever have to recover these VMs during a disaster.
if you are using vSphere replication you might want to take a look at PIT and see if that meet your needs
I want to thank you for solving my problem because of you. I also took the help of online class help but no one gave me a good guide. I understood your point well and my headache ended.
Unfortunately, we're not using vSphere Replication. We're using Recover Point. It lets you do the same point in time snapshots but, it's of a whole volume and not at the VM level. We'd have to revert about a dozen VMs at once. We could do it but, it's a pain.
My solution for now was to put a message in the recovery plan that pauses and says "Hey, go take a snapshot of xxx". Not as elegant as I would have liked but, it works.
You could take advantage of Custom recovery steps in SRM instead of just using the pause dialog.
You would need a helper Windows VM to run powercli commands and use the new-snapshot method