I had a plan for my cluster, but I had a reasonable flaw pointed out to
me, so I am hoping to get advice from you guys.
I wanted to build a simple two-node cluster, running DRBD for shared
storage and a single VM that could float between the nodes for maximum
availability. However, this makes the VM a single point of failure.
The suggestion made to me was to use multipath and run the two nodes as
independent iSCSI targets and then, withing each VM cluster node's dom0,
grab a slice from either iSCSI target and create a RAID1. Alternatively,
maintain the DRBD shared storage and use multipath instead of RAID1.
My concern is that I don't want to have to manipulate the dom0s in the
cluster to rebuild an array if one of the iSCSI nodes dies. Would
multipath (which I've not used before) avoid this? Is there a way to
implement two separate VMs, one on either iSCSI node, that could
abstract away to redundancy from the VM cluster's dom0s?
Basically, I am trying to come up with a cluster config that has the the
fewest (ideally 0) single points of failure while also minimizing the
admin time needed to recover from failure in the iSCSI 2-node cluster.
I'm open to advice and opinions.
Thanks!
Digi
--
Linux-cluster mailing list
Linux-cluster@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/linux-cluster
