vgextend on 11iV3

[System Unix]

Hi,
I have a hp bl870c model in stalled with OS 11iv3. the server is connected to DMX3-24 storage. there are 3 luns in the VG :
/dev/dsk/c2t3d0
/dev/dsk/c2t3d5
/dev/dsk/c2t3d6

when I issue the command vgextend /dev/vgdmx1 /dev/dsk/c4t3d0

I recieve the following error :

Verification of unique LVM disk id on each disk in the volume group
/dev/vgdmx1 failed.
Following are the sets of disks having identical LVM disk id
/dev/dsk/c2t3d5 /dev/dsk/c4t3d0

the 3 paths c4txdy are in CLAIM status and diskinfo gives the information correctly.

any idea ?

thanks,

Reuven

[Torsten. · Accepted solution]

I don't know much about this array, but if it is active/active you should use the new devicefiles like /dev/disk/diskxx, not the legacy. If it is active/passive you cannot/should not use it with 11.31.

Try

# ioscan -m dsf

first to see the new device files.

[Michael Barron]

Make sure those disks aren't just alternate links to the same disk. i,e
/dev/dsk/c2t3d5 is the same as /dev/dsk/c4t3d0, just another path.

How many fibre card / cables do you have running to the SAN? check with the SAN engineer or veify on the SAN that they areor are not alternate links.

Hope this helps

[Tim Nelson]

11.31 includes multi-pathing natively.

You should decide to use it or not to use it. Mixing could get confusing.

I am assuming that c2t3d0 and c4t3d0 are the same lun?

If the above assumption is correct then you do not need to add the other path. Use the /dev/diskxx/diskxx device name. HPUX will manage the paths.

scsimgr and ioscan -m dsf can help you view the relationship and properties.

In 11.31 you can use one or the other ( legacy or agile devices ) so in your case it is legit to do what you are doing, why you are getting an error for an alternate path is weird ( this is ASSUMING that this is an alternate path ).

[System Unix]

Hi Torsten,
thanks for your reply.
the array is active/active. I just wanted to know why there is a change if I use the good old legacy way like in privious versions 11.23,11.11.

thanks,

Reuven

[System Unix]

Hi Michael,
Thanks for your reply.

we have 2 fiber cards on each hosts. these 2 pathes point to the same LUN or device as the server sees it. LUN id is the same. if I use the legacy way both paths should be seen for each device in vgdisplay -v /dev/vgXX

rgds,
Reuven

[System Unix]

Hi Tim,
thanks for your reply.
what do you mean that I should decide whether to use it or not, or mixing ?

c2t3d0 and c4t3d0 are the same lun seen by 2 paths.

can you be more specific please how to use it ? is there anything I should define in scsimgr ?

thanks,

Reuven

[Tim Nelson]

>>what do you mean that I should decide >>whether to use it or not, or mixing ?
I would think it would be very confusing to not use just one method ( there probably is no rule that says you cannot )

>>c2t3d0 and c4t3d0 are the same lun seen by 2 paths.
it is weird that you are getting the error message then

>>can you be more specific please how to use it ? is there anything I should define in scsimgr ?
man scsimgr
e.g.
scsimgr lun_map -D /dev/rdisk/disk3
or
scsimgr get_info -D /dev/rdisk/disk4

[System Unix]

thanks, Tim

[Torsten.]

Be aware you can only use the advantages of the new mass storage stack if you use the new device files.

You can convert your config with a special tool (sorry, can't remember the name at the moment, but it must be in /usr/contrib/bin).

[Torsten.]

Found it: vgdsf

See also

http://docs.hp.com/en/LVMmigration1/LVM_Migration_to_Agile.pdf

[System Unix]

thanks very much Torsten for your good and helpful assistance :-)

rgds,
Reuven

[System Unix]

thanks everybody for your good advise.

Reuven