About 3 weeks ago our 3 node OES Netware 6.5SP7 cluster suddenly went mad (nearly 50 abends since that time). It is currently attached to an EVA 3000 with QL2300 cards. Each node is an HP DL380G3. Errors are across the board and I have few clues, but I do have a couple of them:

1) The abend distribution is lopsided. Node A is less likely to abend than nodes B and C. Node A = 4 abends Node B and C 14 and 15 abends respectively. So, for some reason, A is only ~25% as likely to abend as nodes B and C.

2) I patched all the nodes to SP8 one at at time. After patching the first node, Node C, the cluster stabilized with no abends for almost a week before patching the second node, Node B. When I patched that node, the cluster was destabilized once again. Patching the final node had no effect.

3) A node will always abend at an inconvenient time. ;-) A node can abend even when it has no resources. There's no clear correlation between abends and which node is master.

Anyone have a suggest as to where I might look besides the cluster event log, and CPQIML?

Ken T.