Migration - MSCS along with new SAN Storage

[galibai]

Migration - MSCS along with new SAN Storage

We need to migrate for SAN storage to a new/better SAN. Currently, we are having MSCS cluster on old SAN and we need to migrate this to the new SAN. These MSCS are clusters across box. Currently, they are sitting on ESX 3.5 environment.
I am kind of confused with the steps that needs to be taken. I have a rough idea, as what needs to be done, I would appreciate if someone can check and let me know, if these steps are correct and will not have any issues later.

1) Take the backup of data and VM
2) Take a print screen of all the nic cards settings
3) Remove the existing pointer from the vmfs partiton, which points to the RDM lun on both the VMs
4) Do a storgae vmotion of both the VMs to a common lun on the new storage - This lun will be exclusively for these VMs.
5) Crate a folder under the Lun, where you will keep all your pointers to the RDM luns
6) use vmkfstools -r example - vmkfstools -r '/vmfs/devices/disks/vmha1:2:100:0' rdmname.vmdk
7) Edit the vmx file and add scs1.present="true" - I am not sure, why I am not able to add this from GUI
   a. Add all the necessary parameters so that the pointer will work for the RDM luns
😎 Add all the RDM luns and Quorum via editing vmx file
9) Start the VM and there it will ask you a question - select LSIlogic as your controller card and not Buslogic
10) Once, rebooted assign proper drive letter to the RDM luns. Be careful not to format them, otherwise all the data will be deleted
11) Then, add the nic cards to the VMs
12) Your Cluster is backonline

Somehow, I am not able to make changes in the compatability mode of the Mapped Raw Lun to Physical. It shows Virtual and since, I am using Cluster across box, I want to make both the SCI controller and the Compatability mode to Physical. Any ideas how to do this?

In future, we are going to migrate these ESX boxes to vSphere. We plan to use Round Robin, will that create any problems with MSCS environment ?

Please let me know, if you think there is any problem with this plan of mine.

Thanks

[idle-jam]

Shut down the host and remove the hard  disk/RDM from the virtual machine. Be careful NOT to select “Remove from virtual machine and delete files from disk” when removing the disk. Add the RDM as a new hard disk and select Physical or Virtual as the compatibility mode. After the guest is started , go to disk management (diskmgmt.msc in windows). Round robin will be dependent on the type of SAN storage you have it is advisable to follow what is suited best for that particular model.

[AndreTheGiant]

For compatibility mode, just add the RDM with a new SCSI controller.

About RR, remember that this does not mean more simultaneous paths, only "balanced" paths...

Andre

[rManic]

For RDM luns you should not use RR please refer the following kb

http://kb.vmware.com/selfservice/microsites/search.do?cmd=displayKC&docType=kc&externalId=1010041&sl... 0 171673825

point ot remember if you use storage vmotion the RDM mapping file will be converted into vmdk and all data from the RDM luns will be copied into that vmdk.so use alternate way (unregister the cluster node move to new datastore and register it again).

for cluster across box it is recommented to set scsi adptor to pysical comatability (do this while vms are powered off)

regards

Manick

[galibai]

Thanks a lot for your reply. I will keep these points in mind, while performing the migration.

All, thought I can set the SCSI controller to Physical mode but the compatability mode, I am not able to set it to Physical.

Any idea, what is going wrong here?

Thanks

[rManic]

try the following

Power off both the node and try and try to set

Remove the disk from both the node (do not delete)and readd  and try

regards

Manick

[bulletprooffool]

I have a guide similar to this at:

http://www.get-virtual.info/2011/01/25/replacing-clustered-storage-for-a-sql-cluster-emc-ce-ms-clust...

The MSCS used disk signatures to identify the disk - in the guide I have 3 steps (in your case though I think you can ignore the EMC-CE portion which is for GeoSpan clusters)

basically, what we do is re-create the new disk to look like the old one to the cluster (in other words, prevent having to reconfigure the cluster)

This would of course be a different approach as you are going to need to duplicate the Disk at a SAN level  . . then change the IDs.

Let us know how it goes please.