CA and Vm ware ESX

[Je Thomas]

I am trying to migrate data from EVA8000 to 8400 using CA. I have planned the DR groups.Is there a procedure we need to follow for Luns presented to hosts running VMWare ESXi 3.5 Update 5 and Windows 2003 server.

I am planning to start replication, do a failover, suspend I/O, after the copy is normal suspend the DR, unpresent the LUN from 8000 and present the copied LUN from 8400 with the same LUN #. Either reboot the host or rescan the disks.

Is this the right approach or is there a different way to do this.

Really appretiate your inputs and knowledge.

[Uwe Zessin · Accepted solution]

You must wait for the replication to finish. Then stop the host I/O. Then do the failover.

Take a look into the documentation about resignaturing datastores and re-registering all affected VMs. I have not checked with the latest 3.5 updates, but in previous versions resignaturing was necessary, because the SCSI inquiry string changes when you move from an EVA-8000 to an -8400.

[Je Thomas]

Thanks Uwe for the response.
i was also confused about the resignaturing of VM datastores. do you have any doc or manual where it states how to do the same.

[Je Thomas]

Just to let you know Uwe here we have the same host where we need to present the replicated LUN. Also the hosts are clustered

[Uwe Zessin]

It doesn't matter whether it is the same host. The problem is that some characteristics of the LUN are made part of the file system ID in order to be able to detect snapshots, copies or other changes.

From the ESX command line it is as simple as:
#esxcfg-advcfg -s 1 /LVM/EnableResignature

#esxcfg-rescan vmhba1
#esxcfg-rescan vmhba2
(or what ever you adapters are named)

For ESXi you have to use the VI client or the RCLI.

Don't forget to change the parameter back after the rescan:
#esxcfg-advcfg -s 0 /LVM/EnableResignature


This must only be done on one server - after the resignaturing has happened, you just do a rescan on the other servers and then re-register the VMs.

[Je Thomas]

Uwe is there any side affect to the VM guest hosts that are created from the Datastores.

I will ask the VM Team to consider the resignaturing but they want to know will there be any side effect on the VM guest hosts

[Uwe Zessin]

Side effect? Oh yes!!

When datastores are resignatured, their ID and the path to the VMs changes, e.g.:
/vmfs/volumes/4bc82cc3-66853bf4-5422-0025b3e0fb54/VSA2A/VSA2A.vmx

The "4b...54" part will be different after a resignature. So, it is a good idea to:

- make a list of all datastores and their IDs
-- on each server: # ls -1 /vmfs/volumes

- which VM is stored on which datastore and registered on which server:
-- on each server: # vmware-cmd -l


After the resignature, a datastore will also have a new, temporary name. You have to remove the old name from the vCenter database before you can rename the datastore back to the old one.



I think it is a good idea to try it out before the real move happens:
Create a new datastore and put one or two small VMs on it. Then unpresent the vdisk, rescan and move the vdisk to a new LUN address. Change the parameter for resignature and rescan.

[Je Thomas]

Uwe do i need to change the LUN # while presenting from new storage or do i need to have the same LUN #

[Uwe Zessin]

If you can keep the LUN address and the LUN WWN, there is a certain chance that no resignaturing is necessary. I don't have tried that with ESX 3.5U5 so I can't say for sure.

When you are prepared to resignature anyway, it does not matter if you keep the address/WWN.

[Je Thomas]

Thanks Uwe,i have asked the VM team to have a look at this.

I guess we have no challenges for windows host. I can straight away present the replicated luns and the host will pick it up .
Kindly suggest.