MSA Storage

MSA1000 & FCA2214

Go to solution
marton anka
New Member

MSA1000 & FCA2214


We have the following config:

- MSA1000
- Win2003R2 SQL Server 2000 Active/Passive Cluster, 2 nodes
- Each node has two FCA2214 cards

Nodes are connected in a mesh: Node1/Card1 goes to HUB1/Port1, Node1/Card2 goes to HUB2/Port1, Node2/Card1 goes to HUB1/Port2 and Node2/Card2 goes to HUB2/Port2.

No HP software is installed - there is a Microsoft multipath driver that Windows sets up automatically.

When we bring Node1 online it sees and can access all volumes. When Node2 is subsequently turned on, the volumes are immediately yanked from Node1 (when Node2 is still at the beginning of POST), and will ultimately appear on Node2 once Windows is up. From here on the SQL group (and all volumes) can be freely moved between the two nodes with the cluster administrator software - but every time a passive node is rebooted and it begins its POST, the MSA1000-hosted volumes disappear from the active node. They will ultimately show up on the newly booted node.

We're running with factory defaults on the FCA2214s, and haven't made any config changes to the MSA1000 either.

Can anyone offer any insight into what's going on?
marton anka
New Member

Re: MSA1000 & FCA2214

To clarify the storage configuration:

MSA30 Enclosure
Two MSA1000 controllers for redundancy
Two MSA1000 3-port HUBs

Nodes are HP-DL380s with two FCA2214 each.
Uwe Zessin
Honored Contributor

Re: MSA1000 & FCA2214

> there is a Microsoft multipath driver that Windows sets up automatically.

It is my understanding that Microsoft provides a multipath *framework*, but you do have to get the device-specific module (DSM) from the storage array vendor:

You problem of loosing access, I think is described in this advisory:
marton anka
New Member

Re: MSA1000 & FCA2214


Thanks, this did the trick!

(The 2nd URL is jumbled almost beyond recognition but one can put it together...)

So it's a driver AND a jumper on the FCA2214s. Also - I dug deeper and it turns out that if we'd gone with fibre switches instead of HUBs the jumper problem wouldn't have come up.

James Muell
Valued Contributor

Re: MSA1000 & FCA2214

We had this same problem on our clusters and finally, for windows clusters have gone back to the HP emulex based HBAs which for us are more stable and easier to update firmware using HBAnywhere.