Restarting WebLogic Server without stopping Endeca Server

If you shut down WebLogic Server while the Endeca Server is running, when you restart WebLogic Server, two situations are possible. When Endeca Server is deployed in single-node mode, if you shut down WebLogic Server, Endeca Server starts without delay. If you installed in a cluster mode, Endeca Server needs a limited time period to resolve its startup process.

Note: The remainder of this topic is only relevant if you installed in a cluster mode. If you installed in a single-node mode, Endeca Server starts without delay.

When Endeca Server is deployed in a cluster mode, if you shut down WebLogic Server while the Endeca Server is running, when you restart WebLogic Server, the Endeca Server goes into a transient state. The state resolves automatically after a timeout. To avoid the transient state of the Endeca Server, before you stop the WebLogic Server, stop the Endeca Server application running in it.

The following statements describe this use case in greater detail:

Shutting down and restarting the WebLogic Server without stopping the Endeca Server leads to the Endeca Server entering a transient state. This process continues for a configured period of time until it resolves itself.
The length of the transient state is controlled with the endeca-cluster-coordinator-maxSessionTimeout setting in the EndecaServer.properties file. This setting is balanced for optimal performance of data domains consisting of more than one Endeca Server instance. Oracle does not recommend changing this setting (for example, do not change this setting to 0). If the Endeca Server cluster administrators are considering changing this setting because of their specific network or cluster topology, they should contact Oracle Endeca Support.
While in the transient state, the Endeca Server selects a new leader Dgraph node for the data domain. During this time, its data domain behaves as read-only — it accepts non-updating end-user queries and does not process updating requests. Once the transient state resolves itself, the new leader Dgraph node is elected and the data domain starts accepting updating queries. While the Endeca Server is in the transient state, an informational message is issued. If the transient state cannot be resolved for any reason, an error is issued. If you receive an error, restart the WebLogic Server domain.

For additional information about the Endeca Server cluster behavior (such as, for a description of how enhanced availability is achieved for data domains), see the Oracle Endeca Server Cluster Guide.