I want to know the concerns if I want to do the failover without using SRM, this failover will be for test purposes, I have two clariion storages one on main site and one on DR site, with replication configured between them (in customer site), the customer want to do failover as follows:
1- stop the replication
2- Clone the LUNs on DR site
3- mount the clone LUNs to the ESX server on DR site
4- power on the virtual machines on DR site
is it dowable from VMware side, and what is the concerns here
Thanks for all
It is possible but will be a very manual process - some questions will there be an existing vSphere environment set up in the DR site? If not one will have to be built.
Between steps 3 and 4 you will need to add the VMs to the DR vSphere environment and you will need to reIIP the VMs to matc the IP subnets at the DR site
yes it is doable. However it is much easier with SRM , since it automates the process.
4. rescan HBAs
5. resignature the LUN, cause it will be detected as snapshot and VMFS datastore will not mount automatically (not an issue with NFS)
6 rescan for VMFS
7. add VMs to inventory (register)
8. reconfigure VMs - connect them to appropriate portgroups. If you have RDMs, they will need to be removed and re-added to VMs.
9. power on VMs in the right order
10. re-IP VMs (if needed)
12. manual clean-up: power off VMs, remove from inventory, unmap the LUNs (depending on ESX version, beware of APD issue), remove clones in the array.
In point 7 how I can add VMs to inventory (register), already they are existing according to SRM as those VMs are protected and existing on DR site, how I can power them up after resignature and mount the LUNs
Do I understand correctly that you have SRM in place with protected VMs, however you don't want to run failover via SRM, but manually?
When you add VMs to the inventory at the recovery site, you can specify a different name for them.
Another alternative will be renaming placeholder (shadow) VMs.