Need to change Cluster lock disk on a two node service guard HP-UX cluster.

[Ajay S]

Hello Folks,

WE are migrating storage and I need to change Cluster lock disk on a two node service guard HP-UX cluster.

here is how my setup and plan looks like. i want some export eyes to look at and provide thier advice. if i need to make any changes.

amd i also want to know if i need to halt the cluster for this activity?

--------------------------------------------------

OS version:                         HP-UX 11.31 HA OE

Service Guard version:  11.20

----------------------------------------------------------------------------

Current config related to lock VG (VGLOCK) on both nodes.

Node1   (current active node)

FIRST_CLUSTER_LOCK_VG           /dev/vglock

FIRST_CLUSTER_LOCK_PV /dev/disk/disk24

VOLUME_GROUP           /dev/vglock

Node2

FIRST_CLUSTER_LOCK_VG           /dev/vglock

FIRST_CLUSTER_LOCK_PV /dev/disk/disk24

VOLUME_GROUP           /dev/vglock

----------------------------------------------------------------------------------      

The device file names are same (/dev/disk/disk24) on both the nodes.

Vglock is disabled on both the nodes.

vgdisplay: Volume group not activated.

------------------------------------------------------------------

I presented a new disk for quorum lock but device file names are not consistent for the same LUN.

On node 1:                          /dev/disk/disk83

On node 2                           /dev/disk/disk365

My plan is:

• Create a new VG /dev/vglock1 on node 1 and add /dev/disk/disk83 to it
• Activate vglock1.
• Export in preview mode and copy the map file on second node.
• Import vglock1 on second node (node2)               ( it will import it using device file /dev/disk/disk365)
• Activate vglock1 in exclusive mode on first node (node1)
• Gather cluster information and generate an asci file using cmquerycl.

       Cmquerycl –v –C mycluster.ascii –n node1 –n node2

• Edit the ascii file and make changes as 

Node1   (current active node)

FIRST_CLUSTER_LOCK_VG           /dev/vglock1

FIRST_CLUSTER_LOCK_PV /dev/disk/disk83

VOLUME_GROUP           /dev/vglock1

Node2                                  

FIRST_CLUSTER_LOCK_VG           /dev/vglock1

FIRST_CLUSTER_LOCK_PV /dev/disk/disk365

VOLUME_GROUP           /dev/vglock1

• Run cmcheckconf

Cmcheckconf –v –C mycluster.ascii

• Apply cluster config

Cmapplyconf –v –C mycluster.ascii

• Verify that cluster config using cmviewcl

[Patrick Wallek]

Here's my documentation to change the cluster lock disk:

Modifying Cluster Lock Disk / Cluster Lock VG configuration

1. Locate the current cluster ASCII file. If its whereabouts is unknown, recreate it on one of the nodes using the cmgetconf command.

# cd /etc/cmcluster

# cmgetconf cluster.ascii

2. Insure the cluster.ascii reflects the current hardware and network configuration:

# cmcheckconf -C cluster.ascii

If this command fails, the cluster.ascii may be out-dated.

3 Locate the FIRST_CLUSTER_LOCK_VG and FIRST_CLUSTER_LOCK_PV references in the cluster.ascii file.

If the FIRST_CLUSTER_LOCK_VG reference will be changed, and it is a new volume group for the cluster, insure that each node can see the same set of disks for that volume group. See the /etc/lvmtab on each node.

                Then add the volume group name to the cluster.ascii file. For example:  

VOLUME_GROUP           /dev/<vgname>

4 Update the FIRST_CLUSTER_LOCK_VG reference and/or FIRST_CLUSTER_LOCK_PV references in the cluster.ascii file.

If updating the FIRST_CLUSTER_LOCK_PV - Make sure you do it for ALL nodes in the cluster. There should be a FIRST_CLUSTER_LOCK_PV line for EACH cluster node in the cluster ascii file.

5 When a maintenance window arrives, halt the cluster:

# cmhaltcl -f

6 Turn off cluster aware bit for the VG and then activate the cluster lock VG.

# vgchange -c n <vgname>

# vgchange -a y <vgname>

 7 Use the cmapplyconf command to generate a new cluster binary. This command performs a cmcheckconf validation of the cluster.ascii file, then updates and distributes the cluster binary file to all cluster nodes:

# cmapplyconf -f -C cluster.ascii

Begin cluster verification...

Modifying node <nodeA> in cluster <ClusterName>

Modifying node <nodeB> in cluster <ClusterName>

It should not be necessary to delete the cluster (cmdeleteconf ) prior to changing the cluster lock volume group or disk.

8Deactivate VG and make cluster aware again:

# vgchange -a n <vgname>

# vgchange -c y <vgname>

9 Restart SG

# cmruncl -v -n <nodename1> -n <nodename2>

10 Verify cluster lock configuration with 'cmviewcl -v' command and look for the "Cluster_Lock_LVM" section.

[Ajay S]

Patrick, Thanks for responding back tp me and sharing the information.

so I need to halt the cluster to make this change? I read in one on the forum posts that in SG 11.19 and higher, it can be done online.. just wanted to make sure that.

Thanks Again!

Ajay

[Patrick Wallek]

That very well may be true.  This document was for a customer of mine that has HP-UX 11.11 and SG 11.16 (I think).

I haven't researched for SG 11.19.

[henrygomez]

Hi my friend, did you reemplace you lock disk????