SRM+vSphere Replication+Reprotect takes long time

[JoaquinC]

Hello

This is the scenario, vSphere Replication to replicate VMs from site A to B, VMs protected with SRM and one VM is big

When a recovery plan is executed site B becomes the protected site, then we reprotect the VMs and site A becomes the new recovery site, during the reprotection of the VMs the replication is reversed  and the state of the VM changes to initial sync, this process with VMs that are big takes long time, in my case up to two days.

So, supposing the application hosted in the VMs does not work and we want to move the VMs back to the original protected site (A), The reprotect process fails becuase it times out due the initial sync takes longer that 4 hours and I do not rememer what is the status of the VM in the recovery plan but I belive it becomes green since it is protected and just the initial sync is running. I have not tried but I wonder.

Is is possible to try a recovery plan even thoguht the VM is still running its initial sync after reprotecting it?

I would guess not a supported one but Is there any other option to move the VMs back without waiting for its initial sync to finish?

Thanks in advance, regards

[ashilkrishnan]

Hi JoaquinC​,

Re-protect with vSphere replication is expected to take time as it has to perform checksum operations. It takes time because VR has to read the entire disk for change in data blocks and then replicate any recent changes.

Please look for this question 'What are the replication and synchronization types in VMware vSphere Replication?' under FAQs. (Refer details about Full sync)

vSphere Replication Frequently Asked Questions (FAQ) - The current version of this doc has moved to:...

vSphere Replication FAQ | VMware

I hope that helps

[JoaquinC]

Hi Ashilkrishnan

Thanks for replying, you helped me last time to understand why reprotect takes time so I was trying to not give the focus to process.

My question is more related to what can I do if I need to move the VMs back even if the VMs are still running its initial sync after reprotecting them?

The Planned Recovery Plan I would guess it would time out in the first step since it will try to sync the VMs which  are currently running their initial sync and that takes long time, so it will fail and cancel the process

Would it be valid/possible to run a Disaster Recovery to move the VMs back even if the VMs are still running its initial sync after reprotecting them which ussualy takes up to two days?  How long the process will wait to sync the data before it moves to the next step?

Is there any other option to move the VMs back?

Thank you, regards

[ashilkrishnan · Accepted solution]

Nice to hear from you again.

Once you perform a recovery, these VMs are no longer protected. At this stage, both planned migration/disaster recovery is not an option and a re-protect is required. If the re-protect time is not affordable, then the only option available would be to perform a manual registration outside VR/SRM is required.

Shutdown VM at recovered site

Browse the VM datastore at original site and register the vmx file

By performing this, any recent changes that were performed at the recovered site will not be available.

As a suggestion, we recommend using storage replication for large VMs and utilize the storage array features.

[JoaquinC]

I ended up opening a case and I got the same answer, basically its no possible to run a recovery plan until all the VMs finish their initial sync.

They only workaround as you correctly mentioned is to remove recovery plans, protection groups, replication and then manually shutdown and power VMs but any change done in the recovery site will be lost.

Thanks, regards