Originally posted by: SystemAdmin

Hi,

I am new to this cluster setup. Found that there are some issues with the one cluster server. Now the servers are running as standalone boxes and the applications and databases are started manually and the Cluster VGs are mounted manually. Cluster services are not getting started on both the nodes. Even while doing cluster failover, its not working. Can anyone guide me on this:

I have 2 nodes with rotating topology installed with HACMP 5.1(which is out dated). I am unable to start the cluster services on both the nodes because of below problems:

1. IP non-consistent.. The service ip is pointing to the service ip name and also to the hostname of node1 on primary node and on secondary node it is also pointing to node2. I found that in /etc/hosts the service ip is named as "ServiceIP ServiceIP node name node1" on primary node and on secondary node it is like ServiceIP ServiceIP node name node2".

If I change the entries in /etc/hosts file, will the IP non-consistent error vanishes or do I need to change some other config files. If so, how?

2. Cluster VG major number is not same on both the nodes. Is there any process to change the Cluster VGs major number on node2 to synchronise with node1? Any files to be updated on this?

3. When I checked "/etc/es/sbin/cluster/etc/vg directory", found the files with cluster vg names like..

1. cd /etc/es/sbin/cluster/etc/vg
2. ls -tlr --> this is listing files with the cluster vg names.
3. The files with cluster vg names are having a unique number. What is this number? where will it come from? Even this number is not matching on both the nodes. I have checked similar setups on different clusters and found that these cluster vg files are having similar number in both the nodes.

Can anyone tell me how to synchronize this number on both the nodes and what is the process involved? Is there anything else I need to check before going forward.

Sree...
#AIX-Forum

Originally posted by: SystemAdmin

Hi,

For the first point i don't know how to solve.

For the second point, one solution to resolv your problem is ( considering the node 1 is your reference ):
1/ Make disk readeable and writable for the node 2
2/ Get PVID <-> VG
3/ Export every VG on node 2 whith wrong VG major number
For memorie you can compare VG major between node 1 and node 2 with the command :
node 1 (root) /dev > ls -la _VG_NAME_
crw-rw---- 1 root system 70, 0 Jan 26 13:32 _VG_NAME_
node 2 (root) /dev > ls -la _VG_NAME_
crw-rw---- 1 root system 70, 0 Jan 26 13:32 _VG_NAME_

Number in bold must be the same :-)

3/ And then import every VG whith the following command :
importvg -V MAJOR_NUMBER -y _VG_NAME_ One_hdisk_power

4/ Return in stable situation and test the synchronization between the two nodes.

May this help you.
Regards
#AIX-Forum

Originally posted by: alethad

For your point #1 your /etc/hosts will depend on how you are handling the failover? Are you using takeover with IP aliasing or takeover with IP replacement(service, boot & standby). IP aliasing is the only one supported now so if you don't have it config'd that way you may want to reconsider to save yourself some trouble down the road.

Here's an example of one of my IP aliasing /etc/hosts. Both nodes have to be the same. An you have to have your clhosts file setup also with just the DNS names.

192.168.XXX.210 cpm1en0.unch.unc.edu cpm1en0
192.168.XXX.211 cpm2en0.unch.unc.edu cpm2en0
192.168.XXX.210 cpm1en2.unch.unc.edu cpm1en2
192.168.XXX.211 cpm2en2.unch.unc.edu cpm2en2
204.YY.XXX.210 cpm1admin.unch.unc.edu cpm1admin cpm1
204.YY.XXX.211 cpm2admin.unch.unc.edu cpm2admin cpm2
204.YY.XXX.212 cpmprod.unch.unc.edu cpmprod
204.YY.XXX.213 cpmdev.unch.unc.edu cpmdev
==========================================
clhosts file contents example
cpmprod
cpm1admin
cpm1en0
cpm1en2
cpmdev
cpm2admin
cpm2en0
cpm2en2
By the way you can't synchronize VG numbers. You have to assign that when you create the VG and only is that set at creation.

Good luck

Message was edited by: alethad

Message was edited by: alethad
#AIX-Forum