Problem with scsimgr replace_wwid

[Patrick Briggs]

I need to write a script that clones a LUN on a NetApp FAS-3050c SAN from one HP-UX system to another for backup and testing purposes. I've had it working great on HP-UX 11.11, but now that I'm on 11.31, I'm getting errors because the OS sees a different WWID on the LUN and is blocking access. In order to fix the problem I have to run scsimgr replace_wwid -D /dev/disk/diskN each time I do this.

The problem with running replace_wwid is the /dev/disk/diskN number changes every time I run it which is a scripting nightmare. I need to know what the /dev/disk/diskN is when I go to vgimport the volume so the script won't fail every time it's run.

Question #1: Is there a way to just turn off this requirement to run replace_wwid?

Question #2: If not, is there a way to prevent the /dev/disk/diskN path changing every time it's validated with replace_wwid?

Question #3: If not, can someone suggest a way I can script this so my snapshot clones will work flawlessly everytime without me having to manually go back in and fix it?

[Bill Costigan]

According to the documentation.

scsimgr replace_wwid -D /dev/disk/diskN dsf

should keep the same disk number.

Note: the word 'dsf' after disk name.

I haven't tried this myself.

[Patrick Briggs]

I did what you suggested, I tried adding dsf at the end of the command and it did seem to work. I'm still not completely sure the disk numbers will be consistant from one time to another. If the server is ever rebooted, I could expect the disk numbers to be re-assigned which would cause some scripting problems.

The only other option I can think of doing is a brute force approach writing a script that does on ioscan -m lun and does a scsimgr replace_wwid on every disk ID that is found associated with the NetApp. Once that is done, I can use the NetApp san lun show command to figure out what disk number is associated with the LUN I just mapped to the server. I would get errors with disk numbers that aren't in need of replacing, but if you have any feedback on a better approach, I'm open to suggestions.

[Duncan Edmonstone]

Patrick,

What release of 11.31 are you on:

swlist *-OE

I vaguely remember the inital release of 11.31 having this problem with disks being incremented every time a new copy of a snapshot was produced, but I did think it was a long-time cured... (although maybe it was fixed in the HP disk arrays rather than the OS). Have you checked you're not supposed to have some different settings on your NetApp PAS for 11.31 (many other arrays have a different setting for 11.31 vs. 11.11)

HTH

Duncan

[Patrick Briggs]

HPUX11i-OE-Ent B.11.31.0803 HP-UX Enterprise Operating Environment

I have also installed all the latest patch bundles:
BUNDLE B.2009.08.25 Patch Bundle
FEATURE11i B.11.31.0903.334c Feature Enablement Patches for HP-UX 11i v3, March 2009
HWEnable11i B.11.31.0903.334d Hardware Enablement Patches for HP-UX 11i v3, March 2009
QPKAPPS B.11.31.0909.341a Applications Patches for HP-UX 11i v3, September 2009

Yes, the NetApp SAN settings are different with 11.31. I have the igroup set with ALUA mode turned on.