Join cluster to a group fails with no apparent errors when cluster name matches hostname or a part of FQDN
2 years ago
Originally Published: 2011-09-19
Article Number
000043287
Applies To
RSA Key Manager Appliance 2.5.0.x
RSA Key Manager Appliance 2.7.x
Issue
Join cluster to a group fails with no apparent errors when cluster name matches hostname or a part of FQDN
When a join operation is invoked through RKM Appliance Operations Console (/rkmawa) => Operations tab => Cluster option, the web page shows some messages with a final line saying that a queue job has been made.  There are no further messages or error indications on the cluster being joined to another group.
The primary node of the existing cluster (say ABC) to which the new cluster (say PROD) is being joined to, shows a message similar to the following in a log file /opt/rsa/cluster/logs/join_cluster.<date>.<hour>.log (even though the new cluster name PROD does not exist in /opt/rsa/cluster/master.cluster.info):

Cluster already exists. Not adding PROD.

Cause
The scripts that carry out join operation incorrectly consider a new cluster as existing IF the new cluster name (say PROD) matches with either its own hostname (such as prod.lab.rsa.net) or any of its FQDN domain component (such as rkmapp.prod.rsa.net).
Resolution
Reinstall the new cluster / appliance making sure that the cluster name does not match with its own hostname or any domain component in its FQDN.  Once installed with a new cluster name, it can be joined to the other existing cluster.
Notes
KMA-610