Problems importing a new VG into both halves of a HA Cluster

[columb]

Background

IBM P650's running AIX 5.1, a SAN with 3 x ESS 800 and 2 x DS8107.

Introduced new DS8107 disk arrays into existing SAN. After configuring the SAN switch fabric, the LUN's and the hosts, from both HA Clustered servers ran "cfgmgr -l fscsi0/1". This found both set of the new sharable disks (hdisks and vpaths).

Problem

From one of the HA servers created a new Shared Volume Group (b40ds8100vg) via "smit hacmp". Again using "smit hacmp" I ran a Discover Volume Group Configuration, followed by a Change/Show Resources/Attributes for a Resource Group. Following the selection of the Resource Group, used F4 to select the Volume Group. I see the existing Volume Group and the newly created one. VG brought into the Resource Group ok.

On the second HA server in the Cluster, I again created a new Shared Volume Group (b42ds8100vg) via "smit hacmp" and basically went through the same process as above. However, when trying to add the VG to the Resource Group, used F4 to select the Volume Group. All I see this time is the existing Volume Group and not the newly created one. Anyone any ideas or advice?

Ceci n'est pas une signature
Columb Healy

 

[DSMARWAY]

Hi ,
can you create the failed shared VG that you created on the 2nd HA server on the 1st HA server ?
Becuase you can create all VG's on one server in a HACMP cluster , therefore both servers will have the same Volume groups , then just activate the required ones on either server). If it works from the 1st HA server , then has the second server patched correctly or at the same level as HA server 1
Are both servers got the latest maintenance levels and HACMP fixes ?

 

[khalidaaa]

1. Make sure that the HACMP is installed & all updates are commited:
# lslpp –L ‘cluster.*’
2. On primary node, Make sure that the new VG created is of type enhanced concurrent and the VG Major number is the same as on the secondary node!
To check the next available major number run the following on both nodes:
lvlstmajor
Pick the same empty slot on both nodes for the VG.

Note: you have to varyoff the VG on the primary machine before creating it on the secondary because it is concurrent VG which will be linked to one Node at a time.

I hope this will solve the problem.

regards,
Khalid

 

[columb]

Thanks for your suggestions, We've tried all of them to no avail except khalidaaa's suggestion that all updates are committed. We have a policy of leaving upgrades as 'applied' rather than 'committed'. Surely this shouldn't make a difference?

Ceci n'est pas une signature
Columb Healy

 

[DSMARWAY]

what errors are listed when doing the imports ?

Have you tried importing the volume groups manually ?

 

[ITEng013]

You didn't say what version of HACMP you are running. My experience with AIX 5.1 and HACMP 4.5 is that it cannot properly handle volume groups with vpaths through the smit menus. The volume group must be created from the command line using "mkvg4vp" (you can include all the vpaths in this command). If you have to add vpaths to the volume group in the future, use "extendvg4vp". After you have created the volume group on the first server varyoffvg that volume group. Log in on the other server and execute "importvg". Once you've done that you should be able to use the smit HACMP menus to create logical volumes and file systems.