vSphere Replication when vCenter is down/crashed

[FrenkyB]

I'm investigating to convert our environment from baremetal into a VMware HA cluster, with SRM with vSphere Replication for DR.

I'm trying to find out what will happen with the vSphere replication in one of the following scenarios

1. Site A: vCenter Server Running, Site B: vCenter Server down
2. Site A: vCenter Server down, Site B: vCenter Server down

I'm not really able to find a clear answer on the web. I'm hoping that the vSphere Replication components continue working, i'm not sure if the really need vCenter to work properly after the replication has been setup.

[vbrowncoat · Accepted solution]

vSphere replication is not dependent on vCenter for replication. Is is dependent on the vCenter at the target site for recovery. So in your examples (assuming you are replicating from B>A):

1. Replication will continue. You can restore a VM as needed

2. Replication will continue. You would not be able to recover a VM until vCenter was running.

If you are using SRM, SRM is dependent on vCenter (at the site where SRM is running) as well.

Does this make sense?

[FrenkyB]

Hello GS,

this makes sense, thanks! IIs there any official documenation where this can be found?

Thanks,

Frank

[vbrowncoat]

This is documented in the VR and SRM administration guides.

If you feel this question is answered please mark it so.

Thank you!

[snekkalapudi]

Hi Khalsa, For>>>>>Site A: vCenter Server Running, Site B: vCenter Server down Replication will continue. You can restore a VM as needed I know replication will continue through VR as it is host-based. Can I simply Power-On the place-holder VM ? May be I should find the host to which the place-holder VM is registered to ?

[mvalkanov]

Hi,

At the target site, the replica disks are not associated with the SRM placeholder VM and powering on the placeholder VM won't work.

At actual recovery time, vCenter, VR server and VRMS have to be up and running:

- a failover image is created on top of the replica disks - config files are copied, renamed, disk paths updated

- the SRM placeholder VM is reloaded with the .vmx from the failover image (or if not using SRM, VRMS registers the .vmx from the failover image into vCenter)

- depending on MPIT configuration, disks might be consolidated

If vCenter at the target site is down, you might be able to manually copy the replica base disks, all the delta redolog files on top of them, manually consolidate, manually rename and update the .vmx and other config files. But this is error prone and officially not supported. VR server keeps internal state of which disks are in consistent state (if things have crashed in the middle of receiving delta sync).

Regards,

Martin

[snekkalapudi]

Hi Martin, Thanks for the info.