Storage LUN recognition on IA64 OpenVMS 8.3

[Garrett Cooper]

OpenVMS (AXP) V7.3-2 is capable of seeing/using storage LUNs 0 thru 4095 as long as a valid UDID is provided. OpenVMS I64 V8.3 only seems to recognize LUNs 0 thru 255.

Has OpenVMS I64 been dumbed-down to satisfy some low-end storage system limitations?

[Ian Gray_3]

I don't believe you can have a LUN greater than 255 (its part of the SCSI/Fibre Channel specification). The LUN is part of the disk controller config.

What can be greater than 255 is the OS Unit ID which VMS uses to define the device number e.g. on an EVA a unit ID of 6543 would translate to $1$DGA6542:. Unit IDs on latest releases can be up in the 32K region.

I'd speculate that if you don't set the unit ID specifically on the disk controller then it assumes the LUN assigned to the Vdisk.

[Duncan Morris]

Garrett,

can you indicate what type of hardware is being used for the storage?

Some types of storage limit the number of devices presented per HBA to 254.

See the discussion in

http://forums11.itrc.hp.com/service/forums/questionanswer.do?threadId=1162934

Are you sure that problem lies with the LUN number itself?


Duncan

[Garrett Cooper]

I have NetApp storage configured. I have been using this succcessfully since late 2006. On OpenVMS (AXP) 7.3-2 this is not an issue. NetApp storage will support LUNs up to 4095. OpenVMS (AXP) will accept LUNs up to 4095. Only OpenVMS I64 8.3 (currently being qualified with NetApp storage) seems to exhibit this limitation.

I have an experimental cluster set up with an Alpha DS20E running OpenVMS (AXP) V7.3-2 and an Integrity Server (rx2600) running OpenVMS I64 V8.3. On the storage side, I have many LUNs below lun id 255 mapped to both hosts with absolutely no problems (in fact, I am SAN booted on both hosts). A single test LUN mapped to lun id 260 is seen properly by the OpenVMS (AXP) host, but invisible to the IA64 OpenVMS. This lun is visable, however, to the hardware at the EFI level.

[Duncan Morris]

Garrett,

that sounds pretty comprehensive - and pretty convincing.

Regretfully, I do not have an itanium to check it out myself.

Good hunting.

Duncan

[Rob Leadbeater]

Hi Garrett,

It's been a while since I've used VMS on an Itanium box, but when I did, I seem to recall having to get hold of some different HBAs - the box had previously been used for HP-UX, and the one's in the box weren't supported on VMS.

The thing that surprised me, was that the supported HBAs were QLogic cards, rather than Emulex cards which are used in all of the Alpha servers.

I'll take a guess that the fault lies in the new Qlogic driver code, which will have been specifically written for the IA64 platform...

Hope this helps,

Regards,

Rob

[Garrett Cooper]

I agree. It looks like the driver (Qlogic) code for OpenVMS is the culprit. When using the Qlogic utilities supplied with the ODE (Offline Diagnostic Environment), everything looks normal. The storage array supplied LUNs (even those with LUN numbers over 255) are visible from that level.

[Greg Sollenberger]

I am working with some NetApp storage and have no problem seeing all 12 of my LUN's. How do I provide a valid UDID number?

[Garrett Cooper]

Greg, contact your NetApp sales rep or your SE and he will advise how to do this. As a NetApp engineer, I can't provide your requested info on this forum.