Endeca Server doesn't start

Question

Hi,
I'm using OEID 3.0 on an Oracle Linux virtual machine. Two days ago I've experienced an issue with Endeca Server: when I start the endeca server domain, launching the $ENDECA_DOMAIN/bin/startWebLogic.sh script the domain starts, but in the standard output I can see this error message:
<17-lug-2013 11.03.18 CEST> <Warning> <com.endeca.opmodel.control.ClusterCoordinator> <OES-000056> <OES-000056: ClusterCoordinator process ended. Automatically restarting>
1994415 [Thread-36-SendThread(localhost:2181)] INFO   [ClientCnxn] Opening socket connection to server localhost/127.0.0.1:2181
1994416 [Thread-36-SendThread(localhost:2181)] WARN   [ClientCnxn] Session 0x0 for server null, unexpected error, closing socket connection and attempting reconnect
java.net.ConnectException: Connection refused
  at sun.nio.ch.SocketChannelImpl.checkConnect(Native Method)
  at sun.nio.ch.SocketChannelImpl.finishConnect(SocketChannelImpl.java:599)
  at org.apache.zookeeper.ClientCnxn$SendThread.run(ClientCnxn.java:1143)
1995685 [Thread-36-SendThread(localhost:2181)] INFO   [ClientCnxn] Opening socket connection to server localhost/127.0.0.1:2181
1995685 [Thread-36-SendThread(localhost:2181)] WARN   [ClientCnxn] Session 0x0 for server null, unexpected error, closing socket connection and attempting reconnect
java.net.ConnectException: Connection refused
  at sun.nio.ch.SocketChannelImpl.checkConnect(Native Method)
  at sun.nio.ch.SocketChannelImpl.finishConnect(SocketChannelImpl.java:599)
  at org.apache.zookeeper.ClientCnxn$SendThread.run(ClientCnxn.java:1143)
<17-lug-2013 11.03.22 CEST> <Warning> <com.endeca.opmodel.control.ClusterCoordinator> <OES-000056> <OES-000056: ClusterCoordinator process ended. Automatically restarting>
1997172 [Thread-36-SendThread(localhost:2181)] INFO   [ClientCnxn] Opening socket connection to server localhost/127.0.0.1:2181
1997172 [Thread-36-SendThread(localhost:2181)] WARN   [ClientCnxn] Session 0x0 for server null, unexpected error, closing socket connection and attempting reconnect
This message is repeated endlessy and the server isn't active.
If I try to retrieve the list of data domains using the endeca-cmd command, the result is:
[oracle@testbi endeca-cmd]$ endeca-cmd list-dd
GRAVE: Error while invoking endpoint "http://localhost:7001/endeca-server/ws/manage" from client
endeca-cmd encountered a problem.
caused by:
The Endeca Server returned an error: OES-000075: Caught exception when starting Endeca Server: null
caused by:
OES-000075: Caught exception when starting Endeca Server: null
  OES-000075: Caught exception when starting Endeca Server: null
But if I try to digit in my web browser the url http://192.168.2.16:7001/endeca-server/ws/manage?wsdl I can see that the web services are active and I can display the xml.
Can anyone help me please?
Best regards,
Samuele Scattolini

Answer 1

Hi Samuele,
Another issue might be related to the fact that you are running the deployment on a VM. Specifically, could it be that that the machines you were using aren't able to resolve the names to addresses?  We've seen this on VM machines - they seem to have two names configured, one of which is resolvable externally and one of which is only resolvable locally. If this indeed is the case in your deployment, try the following:
1. Go to $MW_HOME/user_projects/domains/endeca_server_domain/config
2. Check the contents of the EndecaServer.properties file, on all machines on which the Endeca Server is deployed. Specifically, see how the host on which the Cluster Coordinator should be running is configured.
For example, in a three-node Endeca Server cluster, this line may read as in this example:
endeca-cluster-coordinator-hosts=host1,host2,host3
where host1, host2 and host3 are local host names of your machines.
It is possible that if your machine is configured with a fully qualified domain name, such as host1.us.oracle.com, but the EndecaServer.properties file lists this machine as host1, then, due to the fact that it is a VM, the name is not resolvable.
To see if this is the case, try changing this file to use another type of notation (say, switch from FQDN to a short name, or vice versa, depending on what is in your etc/hosts and ipconfig, and see if this works.
Hope this helps,
Here is a doc link about the EndecaServer.properties file: http://docs.oracle.com/cd/E37483_01/server.751/es_install/toc.htm#Changing%20settings%20in%20EndecaServer.properties
It applies to a data domain cluster, but even in a single-node deployment, the Cluster Coordinator is still running, on this same node, and thus its host must be resolvable.
Hope this helps,
Julia