Only I dont remember seeing a vpath nor cfgvpath on HPUX, that makes me think you have some IBM software that does Virtual drive for dynamic load balancing, in such case its the vpath that you should use...
Well I dont know beacuse I havent seen one with HP connected nore do I work with one, but by guessing I would say the SAN is an IBM ESS (shark or whatever its called now) with IBM multipathing software... Just thoughts...
I just wen to look on IBM and found "total storage SAN volume controller" ... and learnt that it also works with HDS9200 and some HP subsystems... I join Dietmar here and believe this should be moved to storage where some competence should be looking and be please to answer...
After some more trouble shooting, I found that it is probably within the SAN. (To those of you that thought this - congrats!!) Trouble shooting consisted of trying to get the disk set up on an AIX system, which we have most of, and where most of our knowledge base is... Thanks to all that contributed.
Apparently all of the LUNs that we were attempting to use had a persistent lock, or SCSI disk reservation bit, on them which prevented the disk from being added to another system.
The command that we used to determine this was "lquerypr". This tool gets installed when SDD is installed on AIX systems. It can be ran from any system with the problem.
The syntax to determine if a persistent lock exists on the disk is:
lquerypr -pv -h /dev/vpath?? <-- if this returns a 5-6 digit hex code then the disk has a persistent lock that needs to be cleared.
If the disk has a persistent lock then the command to clear the lock:
lquerypr -cvh /dev/vpath?? <-- this will clear the lock on the disk and allow it to be used again by other systems.
Strangely enough there is really no documentation about the command other than it gets installed w/SDD. And apparently not on HPUX. I checked.
if u have sdd installed you would be able to run lquerypr and remove the lock. Can't u remove the locak from an aix node after twin-tailing and zoning it to that node ?. I never tried this but u could give it a try.
2105 is ESS is'nt it ? r u using datapath as well ?
Not sure if this will help, but if you are using a SAN and have some type of LUN Masking (where you make LUNs only available to some systems), then HPUX will show some of the disks with a size 0. I believe it has to do with the way the LUN IDs are assigned on the channel. For example, if we assign two disks (SCSI ID 4 and ID 5) but have SCSI ID 2 and 3 assigned to another system, this system shows 2 and 3 as size 0. It's kind of like a placeholder.
I remember when I helped my customer to get his HP-UX systems migrated to EVA we had the same problem (disk size 0). It was 'solved' by installing some patches. One of my colleagues later noted that HP-UX requires 'knowledege' about the disk model - else it will deny access by using size 0 as a placeholder.
Else you might have to install a special device driver that better understands the IBM box - Compaq, now HP, uses one in their multipathing software (Secure Path) for the EVA.
If the reservation is still bothering you, then you may be able to break through that with scsictl. From the man page:
-p pr_clear key device The scsictl command with the -p pr_clear option can be used to clear the persistent reservation from a device. This command can not be used with any other options or commands available for the scsictl command.
This command should be followed by two arguments: key and device. key is a string of characters which was used while setting persistent reservation. This key can be in any of the following two formats:
(a) Text format: can contain any of alphanumeric characters. In this format length of the key should not exceed 8 characters.
(b) Hex format: preceded by 0x or 0X, can contain any of hexadecimal digit. In this format length of the string should not exceed 18 characters in total (including 0x or 0X).
