MSA1000 and MPIO supported configurations

[Alberto Lambertini]

We have 5 Windows servers connected to an MSA1000, 4 of them have a dual-path connection with MPIO. The last server has only a single path connection so we didn't install the MPIO software on it. I'd like to know if this is a supported configuration.
On the "hp StorageWorks SAN Design reference guide" I found that a similar configuration is supported with Secure Path but MPIO isn't mentioned.

[Mahesh Kumar Malik]

Hi Alberto

Yes, it is supported configuration, since MPIO is required for multiple paths to SAN

Regards
Mahesh

[andy steven]

hi

one thing to remember is that if one of your multi path servers has a failed fibre link and the standby controller becomes active your single HBA server will lose its connection, also as failback is not automatic on MPIO version 1.1 basic you with have to do a manual failback to bring controller A back on line.

good luck

[Alberto Lambertini]

Yes Andy, I know this limitation of basic MPIO. I posted this thread because yesterday we had a problem: we did maintenance on two of the multi-path servers, this required rebooting them. First we rebooted a Win2000 server and after 30 minutes we rebooted a Win2003 server. Today I noticed that during this 30 minutes the single-path server lost connectivity to the MSA1000 because a failover was generated after the first reboot and a failback after the second. I did a more in depth analysis and I found that:
- the first server has the HBA in the lower PCI slot number connected to the MSA standby controller (left)
- the second server has the HBA in the lower PCI slot number connected to the MSA active controller (right).
So I have this opinion: when a multi-path server reboots it tries to establish connection to the MSA by its first HBA (the lower PCI slot) starting a failover if the HBA is actually connected to the standby controller. I'm going to verify my opinion with a test ...

[Tv_3]

Hi, Alberto, it is supported, but as you previosly experienced it's not a good idea to do it that way... The things you'ved seen during the maintenance is a normal behavior...

[Alberto Lambertini]

Hi Tv, do you know any HP or Microsoft document that describe this behaviour ?

[Tv_3]

unfortunaly no...

[Uwe Zessin · Accepted solution]

It is mentioned in the Release Notes for Secure Path Workgroup Edition 4.0B (AA-RL4TN-TE - July 2003):
""You must make sure that the active HBA in each node is located in the PCI slot that is activated earlier during server boot than the slot containing the passive HBA. After the active and passive HBAs have been configured in the nodes, the boot order of each HBA can be visually confirmed using the following procedure:

1. Located the ports on the two interconnect switches that are connected to the active and passive HBAs in one of the nodes.
2. Shut down and power off the node.
3. Power up the node.
4. Watch the port LEDs on the interconnect switch to determine which port is
activated first.
5. If this test indicates that the port on the interconnect switch that connects to the active controller is activated last, switch the cable positions and repeat the test to confirm that the configuration has been corrected.
6. If this test confirms that the configuration is correct for this node, repeat this test for the other node(s) in the cluster.

-----
Note: If the active HBA is not correctly configured in the node, the node failover does not occur. If an active node has already failed and failover did not occur to the passive node, the passive node must be shut down and power cycled before the MSA1000 is recognized.
------""