HPUX Service Guard - Issue on FIRST_CLUSTER_LOCK_PV

[dunstan]

hello please advise me.
HPUX version - 11.11
ServiceGuard - 11.15

There are two nodes in the cluster.
One of the VG (vgcommon) is shared. Currently the vgcommon is on the node1 (can not be seen in Node 2)

The issue is, due to the SAN upgrade I have to change the FIRST_CLUSTER_LOCK_PV in the ascii file (but I did not use the cmapplyconf command)
On the node1 when I do the vgdisplay -v vgcommon, this is the output I got.
-----Phisical Volume----
PV Name /dev/dsk/c33t0d0
/dev/dsk/c35t0d0 Alternate Link
/dev/dsk/c37t0d0 Alternate Link

/dev/dsk/c33t0d1
/dev/dsk/c35t0d1 Alternate Link
/dev/dsk/c37t0d1 Alternate Link

/dev/dsk/c33t0d2
/dev/dsk/c35t0d2 Alternate Link
/dev/dsk/c37t0d2 Alternate Link

I have updated the ascii file with FIRST_CLUSTER_LOCK_PV=/dev/dsk/c35t0d0 because only c35t0d0 can be seen on the node1 and node2 using ioscan or powermt. Also diskinfo can see this disk.

But in the node2 syslog.log, i can see this error.
Unexpected error from cluster lock query :- network is unreachable.

1. The alternate disk can be used as FIRST_CLUSTER_LOCK_PV ?

How can I resolve this, Please advise.
Thanks,

[Johnson Punniyalingam]

http://forums13.itrc.hp.com/service/forums/questionanswer.do?admit=109447627+1292339697685+28353475&threadId=1213948

[dunstan]

Hi, Thanks for the info.

Can I use the Alternate Link (/dev/dsk/c35t0d1 Alternate Link) for FIRST_CLUSTER_LOCK_PV ?

Because, this is the disk can be seen on both nodes.

In one of the very old hp forums, some one was told not to use the Alternate Link. Is it correct?

Thanks.

[melvyn burnard]

you cannot use an Alternate Link for a cluster lock device, it mneeds to be the primary link.

Why not try using cmquerycl -v -C testascii -n node1 -n node2 to see what THAT sees as a shared PV fopr the cluster lock disk

[pradeep m]

>>> 1. The alternate disk can be used as FIRST_CLUSTER_LOCK_PV ?
Yes, alternate disk can be used as FIRST_CLUSTER_LOCK_PV.

>>>
Steps to change the cluster lock PV in Serviceguard:

1. create a current Cluster configuration file:
# cmgetconf c1.ascii
In the example above, the cluster Configuration file is called "cl.ascii". Edit your cluster configuration file with the name of the intended cluster lock VG and disk. Note that each node has a reference to
FIRST_CLUSTER_LOCK_PV. Insure the correct /dev/dsk/ reference is used.
(On each node, use 'strings /etc/lvmtab' to locate the path to the lock disk. It may differ between nodes.)

2. Halt the package and the cluster and update the cluster binary file:
# cmhaltcl -f

3. If the VG is already a clustered Volume Group, decluster it first.
Example:
# vgchange -c n -S n /dev/vg01 # if a SGeRAC VG
# vgchange -c n /dev/vg01 # if a standard Serviceguard VG

4. Activate the lock VG:
# vgchange -a y vg01

6) Update the cluster binary file, deactivate the lock VG and start the cluster:
# cmapplyconf -C
# vgchange -a n vg01
# cmruncl -v

Hope this helps.

Regards,
Pradeep