Sunday, March 7, 2010

Important concept to understand with LH

Hello, we just purchased a lefthand P4500 10.8 TB SAN.We created a network raid 1 volume (spanned across the two lefthand nodes)We configured MPIO as documented and it discovered well 4 paths.However, we started to do failover tests by scheduling a power off of one of the nodes (the node on which the virtual manager was running)Of course quorum is lost and access to the volume is lost, but we cannot manage to restore quorum by restarting the virtual manager on the other node. The CMC console asks us to delete first the VM, and to stop it beforehand, but it is marked as offline already.Could you please give me some advice about regaining quorum or some documentations ?Thanks.

Fran Garcia
Mar 2, 2010 09:10:24 GMT Unassigned
Hola Rodrigo :-)You need to configure a FailOver Manager to achieve a resilent cluster. In order to have cluster quorum you need to have at least (n/2)+1 active nodes, and of course that cannot be done with a 2-node cluster.There is a FOM VMWare appliance included in the Lefthand installation CD.
Mar 2, 2010 09:53:31 GMT Unassigned
Hi,HP/LHN recommend a failover manager (FOM) at a third site as mentioned above. With the FOM then your cluster should stay up with no disruptionYou can use the Virtual Manager (VM)but the thing with the VM is NOT to start it on a node -just create it. Then, in your config of two nodes, should one of the nodes fail then you start the VM on the remaining node. This keeps disruption to a minimum. If, as in you case, you have the VM started and it is on the node that crashes you will not be able to start a new VM - as you have found out! Only one VM or FOM allowed per management group.

No comments:

Post a Comment