HP-UX 11i - EMC DMX - over 8 lun issue

[Kenneth Ray_4]

I have a server (recent hpux 11i version) connected via direct fiber to an EMC Symm 8830 and an EMC DMX 1000. When mapping luns from the DMX 1000 I can't see more than 8 luns. However, I have the Symm 8830 connected with a ton of luns and see all of those, but when I go over 8 luns from the DMX I can't see them.
This should be easy to resolve, but I obviously don't know where to look. Seems like it shouldn't be OS patch related because I see all the disks from the other frame.

Anyone have any ideas as to how I might resolve this?

Thanks,

[Steven E. Protter]

I have been waiting two years for a patch that resolves this issue for me with Xiotech Magnitude disk array.

I have seen nothing and been given no date.

You need to contact your HP rep and lean on them hard. In the mean time, you may need to reconigure your lun structure to accomodate this issue.

This issue is complex, and to be sure, you need to open a software call.

SEP

[Kenneth Ray_4]

See, I just don't understand that because I have several other servers connected to XP arrays and like I mentioned earlier connected to an EMC 8830 and on all these servers I see all the luns. Why is it that on this array I don't? Also, my understanding is that this is an issue that was resolved a long time ago.

[Stuart Abramson_2]

You may only configure luns 0-7 on the disk device. Once the luns get to 8, you have to increment the target. If you look at the luns on the 8830, you should see that.

#
# symcfg list -sa 3a -addresses
#

Symmetrix ID: 000184500421 (Local)

Director Device Name Attr Address
---------------------- ----------------------------- ---- --------------
Ident Symbolic Port Sym Physical VBUS TID LUN
------ -------- ---- ---- ----------------------- ---- --- ---

FA-3A 03A 0 0000 /dev/rdsk/c16t0d0 VCM 0 0 000
0001 /dev/rdsk/c16t0d1 0 0 001
0002 /dev/rdsk/c16t0d2 0 0 002
0003 /dev/rdsk/c16t0d3 0 0 003
0004 /dev/rdsk/c16t0d4 0 0 004
0005 /dev/rdsk/c16t0d5 0 0 005
0006 /dev/rdsk/c16t0d6 0 0 006
0007 /dev/rdsk/c16t0d7 0 0 007 <==== lun 7 causes TID to increment
0008 /dev/rdsk/c16t1d0 0 1 000
0009 /dev/rdsk/c16t1d1 0 1 001
000A /dev/rdsk/c16t1d2 0 1 002
000B /dev/rdsk/c16t1d3 0 1 003
000C /dev/rdsk/c16t1d4 0 1 004
000D /dev/rdsk/c16t1d5 0 1 005
000E /dev/rdsk/c16t1d6 0 1 006
000F /dev/rdsk/c16t1d7 0 1 007
0010 /dev/rdsk/c16t2d0 0 2 000
0196 Not Visible 0 2 001
00DC Not Visible 0 2 002
00DD Not Visible 0 2 003
00DE Not Visible 0 2 004
00DF Not Visible 0 2 005
00E0 Not Visible 0 2 006
009E /dev/rdsk/c16t3d0 0 3 000
0085 /dev/rdsk/c16t3d1 0 3 001
0086 /dev/rdsk/c16t3d2 0 3 002
0087 /dev/rdsk/c16t3d3 0 3 003
0088 /dev/rdsk/c16t3d4 0 3 004

This is not a big deal. You can have mor VBUS numbers than the HP server can handle. No problem.

[Stuart Abramson_2]

The patches, from 2001, were to allow non-contiguious luns. You probably have them.

[Kenneth Ray_4]

Stuart,
Understood on the incrementing of the TID. That occurs on all devices upon doing the ioscan and the insf. However, It doesn't occur on these devices from the DMX. That is what I mean when I say that I dont see them.
Again, it seems to work fine for the 8830; do a ioscan and an insf and voila I see all the disks. WHen I do the symcfg for the port that I am connected it lists all the luns that I have assigned to that port. THe ones I can see it displays the device file. The ones that are mapped that I can't see say not visible. SO now I am starting to think could there be some type of masking or something so the host doesn't see it? By the way this environment is directly attached. (No Switch.)
THanks for any more info you can help provide to locate the real problem.

[Geoff Wild]

Sounds like your vcmdb disks are out of date....

Run a syminq, then on all the disks of size: 11520

Run:

fpath refresh -d

We have HP 11i boxes comnnected to DMX 1000's and see all the LUNS we want. Though we are going through a Brocade SAN....

Also, what version of solutions enabler software is installed on the HP boxes?

# symcli

Symmetrix Command Line Interface (SYMCLI) Version V5.2.0.0 (Edit Level: 425)
built with SYMAPI Version V5.2.0.0 (Edit Level: 425)


Rgds...Geoff

[Kenneth Ray_4]

Geoff,
I am unfamiliar with exactly what the vcmdb disks do. I always thought those were used for masking. I think it might be that the ports are not setup for hp-ux. YOur thoughts on that?

[Geoff Wild]

It's possible the ports aren't setup right - you might have to ask EMC on that.

The vcmdb contains the masking info for the host.

BTW - I gave you the old command - instead of fpath, symmask

fpath is the old command.

Rgds...Geoff

[Kenneth Ray_4]

Geoff,
I think you've helped lead me down the right path. Thanks much.

points shall be awarded.

[Stuart Abramson_2]

You don't do Volume Access (probably) if you don't have a switch.

Try "symcfg discover".

Then try "symdev list".

Do "symcfg list -sa "FA" -address" and post it here. (Where "FA" is 03a or something - connected to DMX.)

Stuart

[Kenneth Ray_4]

It was in fact the port settings on the EMC. Needed to set the V flag for the ports. The V flag enables Volume Set Addressing which with HP-UX allows the rollover of the TID like it should so all drives are visible.