MSA2324 dual controllers, HA/MPIO via multiple switches/HBA etc

[Matthew, Millers]

Hi,

We have the following equipment

MSA2324FC (2 x controller)
Cisco MD fabric switches (x2)
Numerous hosts connected in HA toplology to switches

We have the standard HA deployment for MPIO and redundancy. That is, FC.A1 > SW1 ; FC.A2 > SW2 ; FC.B1 > SW1 ; FC.B2 > SW2.

This has been deployed for around 4-6 months now, during deployment we carried out many tests to ensure that the store would stay available in the event of a storage controller or path failure.

This morning we noticed that all the storage was unavailable to all hosts. Looking at the MSA2324FC we notice that the controller A was unavailable, reporting as down.

I would have thought that even with the controller going down, the vDisks/volumes would have been made available via the other controller - this is definently not the case.

The controllers are configured in point-to-point rather than AL. If configured in AL, the ports will not connect. I had a long long support case with HP regarding this issue when the kit was deployed originally.

On the volume presentation, I have presented each volume to each hostID (2 host IDs per machine representing each FC port), these are presented using the same LUN for each seperate volume. Say volume1 > host1/p1/p2 ; host2/p1/p2 (LUN11) | volume 2 host1/p1/p2 ; host2/p1/p2 (LUN12). MPIO on the hosts sees the correct number of paths.

We have around 30 volumes configured in this way, they are all automatically distributed across controllers for ownership.

Now i thought that the expected configuration is possible due to the ULP on these devices?

What do you guys think? Any advice would be appreciated.

Thanks
Matthew

[Matthew, Millers]

Hi,

Apologies to chase this up so quickly, does anyone have any ideas on the above please.

Thaks
Matthew