Cluster node names change

[Pramod M]

Hi Experts,

I was asked to change the hostnames of clusternodes and also cluster name . I used the below procedure for the same.
OS HP-UX 11iv3 HA-OE
# swlist | grep -i serviceguard
B5140BA A.11.31.06 Serviceguard NFS Toolkit
T8687DB A.03.00 HP Serviceguard Cluster File System for RAC with HAOE
PHSS_41535 1.0 Serviceguard A.11.19.00
SGManagerPI B.02.00 HP Serviceguard Manager
SGWBEMProviders A.03.00.00 HP Serviceguard WBEM Providers SD Product
ServiceGuard A.11.19.00 Serviceguard SD Product

1)Stopped cluster.
2)Disabled the autostartup of cluster in /etc/rc.config.d/cmcluster
3) Renamed both hosts by set_parms hostname and rebooted.
4) Edited the /etc/hosts, /etc/hosts.eqiv, /.rhosts, /etc/cmcluster/cmclnodelist
5) Edited the /etc/cmcluster/config.ascii file with new cluster name and nodenames.
6)used cmcheckconf command to check the config file. Then it returned with the below error.
"cmcheckconf: Unable to retrive the existing cluster name
"
7)I removed the ascii file and renamed cmclconfig file and then created the config.ascii file by using cmquerycl.
8)I am not able to see any Multinode package after this. How Can I retrieve the same?

# cmviewcl

CLUSTER STATUS
hq-nqatch up

NODE STATUS STATE
hq-nqatch1 up running
hq-nqatch2 up running

MULTI_NODE_PACKAGES

PACKAGE STATUS STATE AUTO_RUN SYSTEM
SG-CFS-pkg changing changing enabled yes
SG-CFS-DG-1 down halted enabled no
SG-CFS-MP-1 down halted enabled no
SG-CFS-DG-2 down halted enabled no
SG-CFS-MP-2 down halted enabled no
SG-CFS-DG-3 down halted enabled no
SG-CFS-MP-3 down halted enabled no
SG-CFS-DG-4 down halted enabled no
SG-CFS-MP-4 down halted enabled no
SG-CFS-DG-5 down halted enabled no
SG-CFS-MP-5 down halted enabled no


After re-configuration



# cmviewcl

CLUSTER STATUS
hq-qtch down

NODE STATUS STATE
hq-qtch1 down unknown
hq-qtch2 down unknown

Please help

[Matti_Kurkela · Accepted solution]

> 7)I removed the ascii file and renamed cmclconfig file and then created the config.ascii file by using cmquerycl.

In other words, you completely removed the old cluster configuration. You could have had the same effect with the cmdeleteconf command.

Now you must use cmapplyconf to re-apply all package ASCII configuration files too. But before applying the package configuration files, read them to see if they include any node names; change all node names in the package ASCII files to match the new configuration.

MK

[AVV]

Hi,


I understand that it is an oracle active active cluster with CFS. Does a #cmapplyconf -C -P will result to the reflection of the packages ?

[P Arumugavel]

hi,

Your steps seemed to be somewhat complecated.

I could tell you the simple steps may give you clear

Halt the cluster.

Change the node name in all.

Delecte the existed cluster conf file.

Create a new cluster conf template.

Check and apply the conf file.

Start the node and cluster

Modify the application conf file to reflect the changed node name.

Restart the package to get the effect of modification.

Rgds...

[Pramod M]

Hi Arumuga,
I also followed the same steps as you mentioned except the last steps for package. Here in my setup the package is for mounting the cfs filesystems. There are 5 diskgroups and 5 mount points. I just have to mount these filesystems on both nodes. No application packages are configured. here I can see one SG-CFS-pkg.conf file and one SG-CFS-pkg.sh file as attached. Whether running the below command again will do the needful to associate / start the multinode packages with the starting of cluster.
#cmapplyconf -C config.ascii -P SG-CFS-pkg.conf

please advise...

Thanks in advance..

Pramod

[Pramod M]

Hi,

After cmcpplyconf -P SG-CFS-pkg.conf the Multinode package was associated with the cluster. But there is no Diskgroups and Mount points are activated.

# cmviewcl

CLUSTER STATUS
hq-qtch up

NODE STATUS STATE
hq-qtch1 up running
hq-qtch2 up running

MULTI_NODE_PACKAGES

PACKAGE STATUS STATE AUTO_RUN SYSTEM
SG-CFS-pkg up running enabled yes


As shown in my first thread it should show all the DG & MP

What has to do for enabling the same?

Thanks..
Pramod

[Pramod M]

Hi

I can see from the vxdisk list output that, te disk name & diskgroup name are missing.

# vxdisk list
DEVICE TYPE DISK GROUP STATUS
c0t0d0s2 auto:LVM - - LVM
c3t0d1 auto:LVM - - LVM
c3t0d2 auto:cdsdisk - - online shared
c3t0d3 auto:cdsdisk - - online shared
c3t0d4 auto:cdsdisk - - online shared
c3t0d5 auto:cdsdisk - - online shared
c3t0d6 auto:cdsdisk - - online shared
c3t0d7 auto:LVM - - LVM
#

Why this happend?

[Pramod M]

Hi,
Here is an update against this issue.
The CFS Cluster is ready and working after com[pleting the below steps.

# vxdg -C import
# vxdg deport
# vxdg -s import
# cfsdgadm add all=sw
# cfsdgadm activate
# cfsmntadm add (dgname) (vol name) /(mount_point) all=rw
# cfsmount /cfs1




Closing this thread.

Thanks to all...
Pramod