MD3000i - Setting up Host Groups and moving existing Host to new Group

Using ESX hosts,  I created a document earlier this year.    It does not say I shut down my VM's or anything,  so I'd gather I did not do anything on my existing hosts.  But when I was adding the VM's to the redundant host it seems I did shut down the existing VM's, but that may have been because I wanted to bring them up on the new host and make sure they all worked.    ESX stops you from starting the same VM on two or more hosts at the same time.  The remapping of the lun's to the host group are immeadiate and as long as the existing host has already been added to the group, I can't see why there would be any interruption.

Here is a cut and paste of it.

01/10/2009 -  Add redundant ESX host
On the SAN
1. Add the new host and the new host group in one operation when you add the new host.
2. Move the current production host into the host group
3. Re-map the virtual disks to the disk group using the same LUN as it is now.
4. Note:  above operation was weird on allowing the same lun to be mapped so don’t freak out.
5. The Below operations take place on the new ESX host
6. Rescan HBA’s on new host
7. Add storage.  (The storage drives appeared and were all mapped correctly to the luns.
8. Add new virtual machine (custom)
9. Use correct same names as production esx host
10. Chose the correct disk group for the virtual machine
11. Use existing disk for the VM browse the correct disk group for the VDMK file. (c drive)
12. After VM shows up on console, add the second hard disk by once again using existing disk
and browsing the correct virtual disk for the second needed VDMK file. 
Also make sure of any additional network cards to be added for some of our VM’s
13.   Note: I shutdown the VM(s) on the host that were the same as the VM's on the host that I was adding  
14. Start VM(s) on new host and test.

One note about using host groups; the servers in there need to be a cluster (e.g. Microsoft Cluster Services, VMware HA/DRS cluster) or use a cluster aware filesystem (e.g. VMFS, GFS, OFS2, but NOT NTFS/FAT/FAT32).

If you share an NTFS disk on a SAN between 2 (or more) servers that aren't clustered, you'll run into data corruption (could be shortly after doing this, or could take a few minutes/hours/days, but it will happen).

(this is one of the differences between SAN and NAS)

ESX is nice in this regard.  It won't let you really access or start the same VM on different hosts.

With free esxi, you simply map to the same VDMK files and name the VM's the same on the two hosts, and if one goes down, you start the VM's on the other host manually.   Sort of manual Vmotion only you take a short outage.   You don't even need Virtual Center running to achieve it.   Still a ton better if you can at least get foundation to manage the indivudal hosts.

I have tested it thoroughly,  ESX stops you from having access or starting the VM's at the same time.    Certainly don't try to get to your esx vm same stor with a windows host.  That would be bad news.

Thanks to all for the replies.  I am using Enterprise edition of ESX 3.5. We will be doing vmotion with DSR to balance the load with the three hosts.

We are using VMFS file system with a DSR clustor.  I just was wondering if there were any MD3000i issues with moving to a HOST group. Next step would be build the DSR cluster.

Thanks again

No issues.   :)

Sorry for the overload of info there.

PS: Vmotion works like a champ too.  Some failovers have been just a tad on the slow side, not enough to get user complaints,  but hey it's a cheap san.  I never used load balancing, just host / vm failure.