HP Lefthand Multisite cluster

[Tony0001]

I have two sites site A and site S.

Site A has Nodes 1 and 2 and Site S has Nodes 2 and 3.
The four nodes are part of a multisite cluster.

Nodes 1 and 2 are on subnet 172.16.123.0/24 (Primary)
Nodes 2 and 4 are on subnet 172.16.124.0/24
FOM is also on 172.16.123.0/24

When performing tests like taking one node down at a time all works fine, also if i take two nodes down which means a site is down all works fine and by this i mean that the storage volumes presented are still accessible, once reconnected all synchronizes and is back to normal.

I only experience problems when i bring a Node down from each site at the same time.The volumes are lost and marked as Not Available. (Please see attached)

[Davy Priem]

Your volumes are raid 10, so mirror between the site and stripe within the site.

[Dirk Trilsbeek]

in network raid 10, each block ist written to two consecutive nodes. In a multisite cluster, the nodes are sorted alternating over both sides, so that each block is written to a node on each site. But if two consecutive nodes fail, all volumes with data on these nodes will be unavailable.

[Inks]

Lefthand have partner nodes this is especially important in 4 node mult-site cluster.

LUN1 writes block A , Block A is written to Node1 site A and Node3  site S for example, if Block A is dropped from both sites the lun will fail.

Use the availabilty tab under storage nodes to workout the partnerships.Partnerships change when data is striped

Cheers,

Steve