MSA1000 + secure path 4.0c failover

[Rob_222]

Hi all,
I have a query about getting failover working with securepath 4.0c.
My configuration is an HP Bladecentre with 7 x BL20p blades, each with a fibre channel mezzanine card with 1.34 firmware, Windows 2003 server and HBA driver v 9.0.0.13.
These are connected to an MSA1000 with 2 controllers with 4.32 firmware, and 2x 2/8 MSA SAN switches with 3.1.3a firmware.
The securepath agent is installed on all the blades, with the manager installed on one. There are currently 3 logical disks configured in the ACU, with each volume setup in SSP to be available to both HBA's in 3 of the servers.
In secure path manager, both paths from each of the 3 servers to the MSA are shown - one is preferred active, the other is alternate available.
If the fibre cable from one one the servers is pulled, the logical disk cannot be accessed and in securepath manager an error comes up against that server saying the preferred path has failed, however the alternate path still says available when i would expect it to become active.
Does anyone have an idea on what I might be doing wrong ?

[D. v d Coevering]

I don't know how you configured your fabric's, but while reading I think you have a 2 fabric san design.

If this is the case, it works as it should.

An MSA1000 ,although it has 2 controllers, has only one active controller at one time.
So if an active HBA or FC-cable to a server fails, you'll lose te disk's connected to that server.

And a link failure to a server isn't enough to induce an controller failover on the MSA1000.
This only occures when the active link from MSA to switch fails or the active controller fails.

You could (if you want to) manually fail-over
to the other controller via the telnet gui "set controller active" command.

[CA1275161]

As a test can you manually fail over to your secondary path or controller. I would log into secure path and try it as a test, just to make sure both paths are working. I used to work for HP on the storage team and unless things have changed their engineering never supported pulling a cable from the HBA as a valid test even though I have done it on numberous ocassions and it works fine. Remember that all failover on the MSA1000 is controlled by the software not the hardware. Secure Path has to detect a path failure for faliover to take place. I am pretty sure you are not by your description but by any chance are you trying to use a ISL between the switches? Like I said I would first try a manual failover in secure path to make sure the path is working. One more thing. You mentioned that you are using SSP. You can run into strange failover issues when using SSP if you do not set the host mode within the ACU to the appropriate operating system for each LUN. You might want to check that just in case.

[Rob_222]

Thanks for the replies,
I've just installed secure path 4.0c SP2 to fix the issue with PSP 7.2, and it seems to have fixed the failover problem too.
I was able to move luns between controllers ok before with secure path manager SP1 but pulling the active path cable didn't make the alternate path active.
With sp2 installed, pulling the active cable makes the other path active as I would expect and the servers link to the MSA stays up.
I also upgraded the mezzanine card firmware on the BL20P'S from 1.33 to 1.45