topic P4000 failover in StoreVirtual Storage

P4000 failover

madmax81 — Sat, 31 Jul 2010 06:37:34 GMT

Hello,
I have a P4000 cluster with 2 nodes and a failover manager. On each server there are running several VMs. When I do a failover test by powering off the VSA VM on Server1 the storage stays available for all VMs on Server2 and they continue their work without a failure.The Vms on Server1 stop responding. but stay powered on. The ESX on Server1 also shows the storage with status "normal" but it is not browseable.When I power on the VSA VM on Server1 again all VMs on Server1 continue their work without a problem.They don't even have to be restartet.

Is that a normal behavior?

Thanks in advance,
madmax.

Re: P4000 failover

Bryan McMullan — Mon, 02 Aug 2010 10:30:48 GMT

I've seen this behavior with my cluster when testing outage scenerios. If the storage becomes unavailable, basically all VM's go into a paused state until the storage is available again.

This pause state will also occur when the management node that handles the VIP goes down as well.

For your situation though, are you using Network RAID 2 or higher?

Re: P4000 failover

madmax81 — Mon, 02 Aug 2010 12:39:00 GMT

Hello,
I'm using network raid 2. Is it the same behavior with 3 storage nodes?

madmax

Re: P4000 failover

Bryan McMullan — Mon, 02 Aug 2010 13:30:22 GMT

To be honest, I'm a bit confused by your setup. What do you have VSA's when you have a P4000 cluster? Or do you only have VSA's (not actual NSM hardware)? Are the units in a cluster?

Nonetheless, if you have three nodes in a cluster and use NW Raid2, with one node failure...everything should continue to function just fine. You may see a blip in access, where the machines will pause (<30 seconds or so), if the node that fails holds the VIP.

The behavior you have seen is "normal" when the storage is not available.

Re: P4000 failover

madmax81 — Mon, 02 Aug 2010 13:51:20 GMT

The configuration that I have is a HP Vmware Bundle. It contains 2 DL180 G6 server. On each server there is an esx 4.1 running. That esx host the vsa's and the other virtual machines.The esx server itself connects to the datastore that the vsa's provide. On that shared storage the other vm's are hosted.

Hope it is a little bit clearer now. :-)

madmax

Re: P4000 failover

Bryan McMullan — Mon, 02 Aug 2010 16:31:57 GMT

Makes much more sense. Thanks for clarifying.

As for your situation, that sounds familiar to a problem we had a few years ago when we didn't have the load balancing option set on the Server properties of the ISCSI connection in the CMC. You may want to make sure it is set.

Generally, your storage should continue to function if the other VSA's and your FOM continue to work (since you have NW Raid 2). Everything may freeze for a very short time as soon as the VSA goes down, but it should then continue to function when it fails to the still functioning VSA.