Weblogic cluster life cycle

When you run a weblogic cluster you have to deal with several messages BEA-0001xx . It is then difficult to figure out what is the normal behavior of a cluster and the normal messages to get.

Imagine two managed servers ClusterServer1 (listening on port 4001) and ClusterServer2 (port 4002) running on a cluster and sharing heartbeats through the multicast address 237.0.0.1. ClusterServer2 is started first, and then comes ClusterServer1 This is what you’ll get on both logs of the server.

ClusterServer2 logs (simplified)

<Notice> <BEA-000138> <Listening for announcements from cluster Cluster on 237.0.0.1:7001.>
<Notice> <BEA-000133> <Waiting to synchronize with other running members of Cluster.>
<Info> <BEA-000111> <Adding ClusterServer1 with ID -7475S:10.63.17.17:[4001]:clusterdomain:ClusterServer1 to cluster: Cluster view.>
<Notice> <BEA-000102> <Joining cluster Cluster on 237.0.0.1:7001>
<Notice> <BEA-000360> <Server started in RUNNING mode>

ClusterServer1 logs (simplified)

<Notice> <BEA-000138> <Listening for announcements from cluster Cluster on 237.0.0.1:7001.>
<Notice> <BEA-000133> <Waiting to synchronize with other running members of Cluster.>
<Info> <BEA-000111> <Adding ClusterServer1 with ID -7475S:10.63.17.17:[4001]:clusterdomain:ClusterServer1 to cluster: Cluster view.>
<Notice> <BEA-000102> <Joining cluster Cluster on 237.0.0.1:7001>
<Notice> <BEA-000360> <Server started in RUNNING mode>

When you stop one instance of the cluster, you’ll get the following message

One diagram is better than 10 sentences. So here is a state diagram of one instance. Check that there is a difference between the first instance arriving in the cluster, and the second that has to synchronise with the one already running.

clusterstate

Description of these messages

* BEA-000138 : Listening for announcements from cluster name on address:port.
* BEA-000133 : The cluster member is waiting to synchronize cluster information with other cluster members before completing startup and entering running mode
* BEA-000111 : Adding server with ID id to cluster: clusterName view.
* BEA-000142 : When a server boots in a cluster, it detects running members of the cluster and fetches the JNDI tree bindings of the entire cluster.
* BEA-000128 : Updating name in the cluster with new attributes.
* BEA-000102 : This is an informational message to indicate that the server has joined the cluster.
* BEA-000103 : When a server gets suspended, it is no longer a member of cluster clusterName.
* BEA-000112 : Every server in the cluster sends heartbeat messages every 10 seconds over the multicast. Failure to receive such messages for three timeout periods results in removal of the server from the cluster view.

Leave a Reply