VNX5200UNB disk devices keep reappearing

[HEGP]

Hello people,

I've been migrating the LUNs of a 2-node BL860c HP-UX 11.31 cluster from a Clariion CX4-240 storage array to a VNX-5200.
So far so good, new LUNs are seen and migration is performed smoothly using LVM mirroring.

However since I have started mapping LUNs from this new VNX-5200 array to this cluster, I keep getting alerts from HPSIM saying "(WBEM) Lun path belonging to LUN has gone offline."

Looking carefully, these hardware paths correspond to disk with a type of "VNX5200UNB" and the corresponding /dev/disk/diskNNN are not accessible (I/O error)

If I rmsf them by hardware paths (both the agile and legacy disks) they keep reappearing after a while, syslog.log tells me that the sfd deamon has run an insf command to create them back.

I see no obvious error in my configuration, the hosts are registered on the VNX-5200 as "HP non auto-tresspass" with failover mode = 4, which HP tells me are the recommended values.

The EMC Connectivity Guide for HP-UX mentions a few possible causes for this, including tresspassed LUNs. There were two, but I've moved them back to their home SP now and the problem is still there.

This has never happened with the old Clariion CX4-240 array. Clearly I must have done something wrong, but I fail to spot what. The only "unusual" thing I've done is to skip host LUN ID #0  when affecting LUNs to the storage group (I've started at 1) based on some vague warning I had received to avoid using #0.

Any advice greatly appreciated, I can provide whatever extra info asked for.

Alain