Unsolved
This post is more than 5 years old
30 Posts
0
8963
Server 2012 MPIO problems - no PowerPath
We have Server 2012 running MSSQL and need to connect it to a VNX 5200 via FibreChannel. There are four connections to the VNX and currently one QLogic HBA on the server. Before we installed MS MPIO, I saw the typical four DGC LUNZ paths as well as four paths to the one 3TB LUN with host id 250 we have assigned to this storage group.
The server is currently using an internal HP Smart Array controller, which is where all of the SQL data currently resides. Now that MPIO is installed and, I thought, configured, I see the following:
In disk management, there are 4 unknown disks (offline) and one 3TB disk (online). This disk shows as having one 2TB Healthy partition with 1TB unallocated. The 2TB partition was not created by us and also shows as being 'GPT Protective Partition'.
In server manager under disks, I see 4 offline disks, the DGC LUNZ luns. I also see the 3TB disk listed as being partitioned as MBR and name DGC VRAID Multi-Path Disk Device.
I am unable to remove the 2TB partition, reset the disk, or otherwise change anything. Every operation other than listing partitions, etc. returns 'The device is not ready'.
No, we do not have PowerPath. I suppose I could install it since we only have one card currently. But, MPIO should work. I know that MBR is the reason the 2TB might be showing up. Anyway, I am not sure what I am doing wrong.
iop2go
54 Posts
0
April 1st, 2015 05:00
I know this is going to sound very "windowish", but have you rebooted ?
I could not make 2012 r2 behave correctly with MPIO before I installed the MPIO services, then rebooted.
mlott-gie.com
30 Posts
0
April 1st, 2015 06:00
Oh yes. After installing MPIO and then after adding the device in MPIO Devices we were asked to and did reboot. FWIW, the following shows in that list (Device Hardware ID):
DGC LUNZ
DGC VRAID
HP LOGICAL VOLUME
Vendor 8Product 16
kelleg
4.5K Posts
0
April 2nd, 2015 10:00
In the Storage Group for the hosts, you need to add a LUN that has the HLU set to zero - this will prevent the array from sending a LUNZ to the host (search on "LUNZ" in the Knownledgebase).
Also see: mydocuments.emc.com - select Attaching a new Server to get an installation guide - it has more details about connecting a new server to an array.
glen
mlott-gie.com
30 Posts
0
April 2nd, 2015 15:00
Does having a LUNID 0 present fix my MPIO setup? I'm just wondering whether that's the magic.
kelleg
4.5K Posts
0
April 3rd, 2015 08:00
Yes - just create a small LUN and add it to the Storage Group and before you Apply, click in the column called Host ID - you should get a drop down - select 0 - that will make that LUN HLU 0 and remove the LUNZ. You will need to reboot the host(s) in the Storage Group - you might want to name the LUN "DONOT REMOVE LUNZ" or something like that to prevent someone from removing this LUN by accident.
glen
mlott-gie.com
30 Posts
0
April 3rd, 2015 09:00
Thanks, I will give that a shot and reboot Easter morning.
mlott-gie.com
30 Posts
0
April 8th, 2015 16:00
Still not working. GPT Protective Partition
kelleg
4.5K Posts
0
April 10th, 2015 09:00
You added a new LUN to the Storage Group and select that new LUN as HLU 0. Is that correct>
When you rebooted the host, did the new LUN show up? I'm not sure but you probably need to format that new LUN.
With an HLU 0 (zero) LUN in the Storage Group, you should not see an LUNZ from the host. If you are, then something is probably wrong on the host.
glen
mlott-gie.com
30 Posts
0
April 13th, 2015 08:00
After another reboot, caused by a crash trying to format one of the LUNs, LUNID 250 does appear to be working finally. It's listed as being in Round Robin with Subset mode, showing 2 paths optimized and 2 non-optimized. I don't know if that's the right choice, but I don't see simple Round Robin in the list anymore.
mlott-gie.com
30 Posts
0
April 19th, 2015 07:00
For what it's worth, I see the same mode on an iSCSI-connected Windows host in MPIO - Round Robin with Subset.
kelleg
4.5K Posts
1
April 22nd, 2015 13:00
For all VNX arrays and ESX 5.x and higher or Windows 2008 and higher the failover mode for the hosts must be set to 4 (ALUA) and the for ESX the PSP must be Round Robin only. With four total paths from the host to the array, you should see two as active and two as standby. The two active paths should be one from each HBA - same for the standby paths.
glen