rx6600/HP-UX 11.31 - Panic'ing with "Unaligned reference fault in kernel"

[Alzhy]

HP still clueless.

Anyone experienced this kind of malaise yet?

It's been pretty stable until recently when we added 4 LTO-4 Drives.

The server has 8x4Gbit FC-HBAs. 4-HBAs are used for SAN Disk presentations, the other 4 for LTO-4 FC Tapes. It had 10 LTO-4 TapeDrives presented to it previously.

I know -- that is NutZ! 14 LTO-Drives on 4x4Gbit FC-HBAs? Well you see - this server is running NBU and it's designed to not use all 14 drives at once. The 14 Drives are SSO'd (shared) amongst SAN Media Clients -- in a NetBackup Configuration).


Other things:

It runs VxVM 5.0. It is connected to HDS Thunder, HP EVA and HP XP Storage arrays.

[TTr]

Other than checking the zones as I replied in your previous post, check if there are HP-UX patches that you need to apply. This smells like a segment violation panic.
PS. We have several servers with disk SAN and tape SAN access like yours. Nothing wrong there.

[Steven E. Protter]

Shalom,

Patches or possibly a scsi panic. With all those devices connected, maybe one of them is not terminated properly.

I know fiber is different, but it works and acts like scsi.

Check the 4 new LTO-4 drives.

SEP

[Alzhy]

We're practially all recent revs of patches/drivers, etc.

When this thingy fiest staretd, the panic stck trace matched hose mentioned in a very recent patch for 11.31 estape. That was applied but to no avail.

Subsequent updates to CommonIO and SAS (SERIAL-00) were also done. Same thing.

Firmware was updated on the LTO Drives just to elimnate that as a possbibiliy. No dice too.

Only thing remaining is try movng back to just 10 LTO drives.

OR,

Try applying the week old VxVM/VxFS 5.0 GA patches.

My client's env is BIG, Different and the mix and match of HW is probably unparalleled. We did get a FULL dump this time and hopefully that will allow HP some clue to a possible "instrumented" kernel to dig deep into the panics as they did an 11.11 Env some years back.

[Dennis Handly]

Are you already in contact with the Response Center?

[Alzhy]

FYI -- HP finally discovered the flaw was with a STREAMS/Networking driver -- a custom PHNE patch that will hopefully be GA soon was issued to us and the problem's gone.