cmgetconf reporting error messages

[Prashant_14]

All,

Please look in file I am attaching.

There are few error messages like xxxx disk does not have a ID on xxxx node or disk label.

And also something like :

Disks which do not have IDs cannot be included in the topology description.

The volume group /dev/vgp_elg_d8 is activated on more than one node:

mozart

elgar

I want to know if these are normal messages in a cluster configuration.

Or

What is the reason I am getting these messages, what I will need to do to get these sorted?

This is a three node cluster, having
B3935BA A.11.09 MC / Service Guard

and EMC 3430 disk array installed.

Please revert back for any information needed.

Best regards
Prashant

[Duncan Edmonstone]

Prashant,

You haven't attached your file...

That said, these messages are quite normal (typically they are referencing CD-ROM devices)

Post your file, though and we can check its nothing to worry about.

HTH

Duncan

[Duncan Edmonstone]

Oops,

I missed this bit the first time round:

The volume group /dev/vgp_elg_d8 is activated on more than one node:

mozart

elgar


This isn't good, and could lead to data corruption if you already have something in your file systems on these volume groups. On the system you aren't running the cmgetconf/cmcheckconf/cmapplyconf commands on, deactivate the volume group (vgchange -a n vgp_elg_d8)

HTH

Duncan

[Prashant_14]

Sorry,

here is file!!

Regards
Prashant

[Zigor Buruaga]

Hi,

Check this link:

http://bizforums.itrc.hp.com/cm/QuestionAnswer/1,,0x7c7787dc4d7dd5118ff00090279cd0f9,00.html

Kind regards,
Zigor

[Zigor Buruaga]

Hi again,

About the volume group activated on more than one node, what about using "exclusive" mode.
In "cmcluster.ascii" file you will see something like:

# List of cluster aware Volume Groups. These volume groups will
# be used by package applications via the vgchange -a e command.
# For example:
# VOLUME_GROUP /dev/vgdatabase.
# VOLUME_GROUP /dev/vg02.

You could add the line: VOLUME_GROUP /dev/vgp_elg_d8
and then run cmapplyconf.

Hope this helps.
Kind regards,
Zigor

[melvyn burnard]

As per your file attachment, they are simply Warnings about the fact that certain disks are either not pvcreated and hterefoe cannot be used in the cluster until you do a pvcreate on them, or htey have been setup for Vxvm useage, which means you cannot use htem unlesss you have the patch installed on HP-UX11.11 ONLY to allow you to use VxVm disk groups in ServiceGuard.
As for hte message about the vg being activated on more htan one node, it speaks for itself. De-activate them on one of hte nodes, and hten try again.

[Sundar_7]

Hi Prashant,

# echo "0x2008?4D" | adb /dev/dsk/c0t2d0

Execute the above command and see if it returns something other than 0s in all the fields.

Thanks,

Sundar.

[Prashant_14]

Sundar,

What is that command about? What's expected from that command in result? Hope you understand that I am in production environment.

Regards
Prashant

[Sridhar Bhaskarla]

Hi Prashant,

If your disksystem is HDS/XP or EMC you may see some place holder LUNs and they will be of size 0. You can safely ignore these messages.

However, the real error that is preventing you from applying the configuration is about the volume group that is activated on multiple nodes.

Go to the rest of the nodes and issue the command "vgchange -a n /dev/vgp_elg_d8". Activate all the volume groups that are part of serviceguard on the current node "vgchange -a y" and then apply the configuration.

-Sri

[Tricia Sambrano]

Prashant,

Long term though you need to fix the fact that vgp_elg_d8 can be activated on more than one node at a time.

Next change management you need to make that volume group cluster aware by taking down the applicable package and performing
vgchange -c y vgp_elg_d8

This will prevent the volume group from being activated non-exclusively.

Be sure to update the packge control script so that it is activated
vgchange -a e
not
vgchange -a y

Have fun!
Trish