SRM site to site failover, 1 large datastore on each side, how to get it to work?

[gtprojectx]

Reference this question: http://communities.vmware.com/message/1436162#1436162

I have 2 sites. Each configured with their own vSphere server. Each site has production VMs. I want each each to be able to failover to each other. Each site has 1 datastore configured to the maximum capacity of the storage with running VMs in it.

I configured Site A to failover to Site B perfectly. Site B shows the placeholder VMs. The problem comes when trying to configure Site B to failover to Site A. When I create my protection group it includes the placeholder VMs. When trying to configure the VMs for protection they time out everytime. After timing out, the "Edit Virtual Machine Properties" dialog box pops up anyways. I complete the selections of how to protect the VM, however that times out as well.

I believe it is because there are already protected VMs, albeit placeholders, in this protection group (at Site B). If that is not the case please let me know this is a supported configuration and suggest further steps to make this work.

Thanks

[admin]

The general best practice is to have a very small vmfs lun created for the placeholder VMs to keep them seperate from any production VMs. We use a 4GB VMFS LUN (but could be smaller depending on the number of VMs) for the "shadow VMs". That LUN is not replicated.

I honestly have not tried to place a VMFS containing the shadow LUNs into a protection group, so I'm not sure if your results are consistent with that, but I would go for the best practice.

-alex

[gtprojectx]

Can that small placeholder datastore for the VMs be localstorage on one of the ESX hosts?

[admin]

Good question, I have not tested that scenario, but I would think it's not supported. If you have multiple ESX hosts the others will not be able to see the placeholder VMs and you may be worse off in that only one VM can be recovered at a time on that single ESX host.

I would try and get a small VMFS created on shared storage just for these guys....

-alex

[gtprojectx]

That is my predicament. Based on the other thread, all shared storage has already been allocated.

[admin]

Ah, i did not see that. Well...I don't believe it is supported (although I have nothing to back that up, a call to support would clear that up), but you could try spreading the placeholders across all of the esx hosts on local storage. Don't know if it'll even work, but sounds like you're in a bind and it might be worth a shot to test out. My test envronment is down at the moment so I can't even test that scenario.

[gtprojectx]

Ok. I moved all the placeholders to local storage. I then blew away my faulty Protection Group and created a new one. Of course it found and added the correct VMs at Site B. However during the Protect Virtual Machine task it will time out and leaves the status of the VMs as "Not Configured".

If I try to protect them individually, it will time out. It will then pop up the dialog box as it should to let me configure it for protection. I select all the settings I want and apply it, but that times out as well and the VM stays "Not Configured"

Thoughts on where to look to see what's going wrong? This was the original problem that I thought was being caused by having the placeholders in the protection group. Now that the placeholders are not in the protection group I'm a little stumped on what to do next.

[admin]

You can look in the logs on the SRM server c:\Documents and Settings\All Users\Application Data\VMware\VMware Site Recovery Manager\Logs

It's possible that local storage just won't work...

[gtprojectx]

Where would they be in server 2008?

[admin]

On any platform you can look under Programs > VMware > VMware Site Recovery Manager for the log generation tool. Or look for "SRM log files" in the SRM admin guide.

-alex

[admin]

Hi guys,

I would suggest that you use the log bundle generation tool to collect the logs. Sometimes customer just send us the current log and than we have to ask for the other logs to get the full picture. This tool is in the VMware program folder.

As well, you do need a shared LUN for the shadow VM placeholders. This is due to the way we register VM's and start them. With SRM 4 we would only use the ESX host that had the local storage with the placeholders and we would ignore the other hosts. So for that reason the shared LUN accessible by all of the hosts at the recovery site is best.

Michael

[gtprojectx]

I generated the log bundle using the tool. Is this something I should attach to this thread or send to someone or try to parse myself?

Thanks

[admin]

You can submit that log bundle, or the one from each side, to VMware Support and they will be much more able to help you with your problem. I was not suggesting you submit the log bundle to this thread, or support, just that it is better to gather the log with the tool, rather than grab files individually.

Michael