MSA MPIO Disk Not Initialized

[Eric Distad]

I'm trying to make sense of a weird problem.

I just installed a DL585. We have a MSA1000 with redundant switches from the High Availability Kit. I installed the correct drivers and the MPIO and everything looked fine. I have 2 LUNs assigned via SSP to the server and they are having no problems.

At one point I powered down the server, as I have several times in the past, and I received the message from Insight Manager that the Redundant Controller was Active. When I logged in one of the LUNs was missing. I had an F: drive but not G: as I had before.

I started doing a process of elimination. I determined that if the path to the MSA is through the newer of the 2 switches - the one that came with the HA Kit - that everything works fine. If it is plugged in through the older switch - SAN Switch 2/8 - The volumes show up in Disk Management, but they are not Initialized.

Can I Initialize these disks without harming the data on the Drive? Could the difference between the 2 SAN Switches cause this or is this a Disk Signature issue?

None of the other servers on the SAN seem to have an issue with this. Is it because the new server is running Windows 2003 64-bit (unlike the other servers)?

I'm trying to make sure that when I re-setup and install MPIO, that it will work properly.

Thanks for any help you can offer!

[Bill Costigan]

I would not initialize the disk. There is only 1 LUN so anything you do to it through one path also effects what is seen through the other path.

If you have two HBAs in the server then you should see them as two hosts in the MSA, one for each HBA. Is it possible that you only gave one of the HBAs access to the LUN through the SSP?

[Eric Distad]

I verified SSP is or at least apperas to be set correctly for both hosts on the MSA.

If SSP had not been set correctly, would both disk still show up in Disk Management when the server is connected through the older switch? When I've used SSP, it's appears to be an On or Off sort of setting. Are there other settings/parameters I'm missing?

[Bill Costigan]

You're right it would not have been seen. However, within SSP there is a setting for OS type for each HBA. If the second HBA was set for the wrong OS type, you might be able to see the disk but something might break during the access.

[Eric Distad]

Ok. That makes sense. The weird thing is both Hosts in SSP are set to Default (As are all the other hosts on the other servers). Should I switch them both to Windows?

[Bill Costigan · Accepted solution]

I'm guessing the default is windows. But I would change it to windows just to be sure.

Also, did you do a rescan for disks in windows disk manager? It might have some old info.

Oh, one more thing - double check that you have the correct version of MPIO installed. There are a number of versions that cover different array types.

I assume the MSA is active-passive which means that only one controller can be used at a time. It would make sense that the LUNs would be seen but could not be accessed via the passive controller until one of the hosts instruct it to "take-over"

If the MPIO software is for an Active-Active array, it might not know enough to send the command to switch the control.

[Eric Distad]

Thanks Bill. I did set the SSP to Windows.

As noted in the original post, I actually uninstalled MPIO in order to test this out.

The problem definately seems to be switch related. No matter which HBA I use, if it is routed through the older switch, it shows connected, the disks show up in Disk Management, but they are listed as uninitialized.

As a note, this server isn't in production yet. I moved the data from the affected SAN Drives, by connecting to the newer switch, and then switched back to the old switch and tried to initialize, to see what would happen, since the drives were empty. It gave me an error saying that it couldn't initialzie the drives and consult the System Event Log for more details. The catch was there were no new entries in the System Event Log.

I'm looking at updating the firmware on the switch... though I'm not sure how that would fix this. The switch in questions - the San Switch 2/8 has firmware 3.1.0 on it. I know there is 3.2.1b out there, but I couldn't find any references to fixing support for newer HBAs or anything like that.