How To
Summary
There could be situations where the SSP cluster could get created that uses other network interface instead of using one dedicated for SSP.
Configuration sample.
1. Management - management local area network
2. Backup - back up and restore network
3. SSP - shared storage pool cluster network
Objective
How to check and correct hostname and IP address for SSP cluster configuration.
Changes are not dynamic hence it is highly recommended to schedule down time for these changes.
Typical error which may occur as an example shown below is when adding a node to SSP cluster and incorrect SSP network interface is not selected.
"Error: ADD_NODE failed for <VIO#2hostname>.
Hostname resolution of <VIO#2hostname> on initiator and target node do not match".
Environment
VIO Server 2.2 and 3.0
Current supported release: https://www.ibm.com/support/pages/powervm-vios-lifecycle-information
Steps
Checking what interface is currently selected and used.
1) $ oem_setup_env
2) cd /usr/lib/cluster
3) ./clras dumprepos -v
4) Search for gw.ip and that is what is being used and can be changed for customers configuration.
Additional Information
Steps To Modify
1) When VIO has more then one IP for various goals and need to use only the IP for SSP we can use "chcluster" command to modify this.
This has to be done from current node that is running state shown below.
To Check cluster status use below command.
$ cluster -status -clustername [cluster-name] -verbose
$ cluster -status -clustername [cluster-name] -verbose
All below should show for the node from where the change is being made:
Node Name: Node_AA
Node State: OK
Node Repos State: OK
Node Upgrade Status: ON_LEVEL
Node Repos State: OK
Node Upgrade Status: ON_LEVEL
2) Then apply below steps to add the new interface.
$ oem_setup_env
# chcluster -m <node1>{+cle_ip=<SSP_IP_node1>}
...
# chcluster -m <node10>{+cle_ip=<SSP_IP_node10>}
...
# chcluster -m <node10>{+cle_ip=<SSP_IP_node10>}
3) Now run "lscluster -m " and make sure all nodes are UP and cluster is healthy.
# /usr/lib/cluster/clras dumprepos -v (verify cle_ip under each node)
4) Finally, apply below steps to remove old the interface.
# chcluster -m <node1>{-cle_ip=<MGT_IP_node1>}
...
# chcluster -m <node10>{-cle_ip=<MGT_IP_node10>}
...
# chcluster -m <node10>{-cle_ip=<MGT_IP_node10>}
Additional Steps to change hostname:
1) All nodes need to be resolved locally
# cat /etc/netsvc.conf
hosts=local4,bind4
# cat /etc/netsvc.conf
hosts=local4,bind4
2) Stop the node whose hostname is being changed from another node which is UP (say A)
$ clstartstop -stop -m oldname_hostname
3) On the node whose hostname is being change update hostname
# chdev -l inet0 -a hostname=newname
inet0 changed
4) From a node which is UP, change the name of the stopped node.
# /usr/sbin/chcluster -m oldname{cle_hostname=newname}
# /usr/sbin/chcluster -m oldname{cle_hostname=newname}
5) Remove the old hostname entry from /etc/hosts from all nodes
6) Restart cluster services
$ clstartstop -start -m newnamehostname
Contact: rajpat@us.ibm.com
Document Location
Worldwide
[{"Type":"MASTER","Line of Business":{"code":"LOB57","label":"Power"},"Business Unit":{"code":"BU058","label":"IBM Infrastructure w\/TPS"},"Product":{"code":"SSPHKW","label":"PowerVM Virtual I\/O Server"},"ARM Category":[{"code":"a8m50000000L0FXAA0","label":"PowerVM VIOS"}],"ARM Case Number":"TS010221519","Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"All Versions"}]
Was this topic helpful?
Document Information
Modified date:
30 August 2022
UID
ibm16611265