After restarting one or both nodes of an HA cluster, the node fails to start successfully.
Looking in the Foglight Management Server logs, the following errors are noted during startup:
---
2015-07-16 16:13:39.777 ERROR [main] org.jgroups.protocols.TCPPING - failed creating initial list of hosts
java.net.UnknownHostException: yourHAFMShostname
2015-07-16 16:13:39.781 WARN [main] org.jboss.system.ServiceController - Problem creating service jboss:service=Partition8fb7ad42-83e8-4c5f-8966-f844272ea201
org.jgroups.ChannelException: unable to setup the protocol stack
---
Or:
---
Caused by: java.lang.Exception: Conversion of initial_hosts in TCPPING with original property value YOUR_FMS_HOST[7800],otherhost[7800] failed
...
Caused by: java.net.UnknownHostException: otherhost: Name or service not known
---
Communication issues between the HA nodes.
Likely because one server is not able to ping the other using the hostname as defined within cluster-service.xml Can be caused by DNS changes or other network related issues.
Ensure that primary FMS node can ping secondary, and vice versa using the hostnames found in $FMS_HOME/config/cluster-service.xml
NOTE: The filename has changed in Foglight 5.9.1 to $FMS_HOME/config/jgroups-config.xml
More regarding High Availability tuning can be found here.
