cmquerycl misses the Poss. Cluster Lock devices

[Andreas Mueller_1]

L class Server running HP 11.00, 2 nodes in cluster:

Somewhat baffling, when I issue the
cmquerycl -v -C /etc/cmcluster/apspitt.config -n apspitt0 -n apspitt1 command, the system will not display my

Possible Cluster Lock Devices:

/dev/dsk/c4t0d0 /dev/vgmsplib 30 seconds

/dev/dsk/c6t0d0 /dev/vgmsplib 30 seconds

/dev/dsk/c6t1d0 /dev/vghome 30 seconds

/dev/dsk/c4t1d0 /dev/vghome 30 seconds

I cut and pasted this info from another cluster I had set up previously, but it is missing on the machine I'm working on now. pvdisplay & vgdisplay indicate that pv and vg are setup properly.

Any help is GREATLY appreciated.
Andy

[melvyn burnard · Accepted solution]

are you sure the VG is named the same on bothe nodes?
Can you activate the VG on each node? are the Minor numbers of the VG the same?
It sounds like you have an LVM config problem.

[Mark van Hassel]

Hi,

You have to set the lock devices in etc/cmcluster/apspitt.config :

FIRST_CLUSTER_LOCK_VG /dev/vgname

and then for each node (depending on the device address):
FIRST_CLUSTER_LOCK_PV /dev/dsk/cXtYdZ

which is a disk in the above specified vg.

Hope this helps

[Andreas Mueller_1]

Mark,
it is my understanding that the system should have found my cluster lock devices, rather than me inputting them into the apspitt.config file. Just like cmquerycl finds and populates most everything else (lan devices, etc). But I will do so and take it from there.

[melvyn burnard]

If your lvm config is correct, then you are correct. the probe done at cmquerycl time should identify the shared vg(s) and pick the first available shared VG as the CL VG, and the first available matching disk as the cluster lock disk, and put the appropriate entries into your ascii file.
One other possible suggestion is to try to have the shared vg activated on the node you are issuing the cmquerycl

[Andreas Mueller_1]

Mark & Melvyn:
I edited the apspitt.config file per Mark's suggestion.

When running cmcheckconf I'm seeing it had problems accessing my cluster lock volume group on the other node. Low and below the minor number weren't the same.

thanks for your help ....

Andy