Deactivating one Message processor in cluster which has two and shutting down the node, deactivate other MP also in another node.

[Nirothipan]

Description: After deactivating one message processor in the cluster, shut down one node and another message processor is also getting deactivated.

Affected Product Version: EI 620 WUM update pack

Steps to reproduce:

• Create a Non-worker manager cluster with 2 nodes.
• Configure the hazelcast properties file as in [1].
• Create two message processors ( say A and B ) and make sure they are working properly. ( I tried with one JDBC and one InMemory )
• Deactivate one message processor ( say A ).
• Shut down the one node.

OutCome: From the Management console of another node, we could observe that both message processors are deactivated.

Expected: That message processor we deactivated is still not active in another node and other processor is active.

[1] https://docs.wso2.com/display/EI620/Clustering+the+ESB+Profile#ClusteringtheESBProfile-ConfiguringHazelcastproperties

Related Issues: https://github.com/wso2/product-ei/issues/2358 https://github.com/wso2/product-ei/issues/849

[Nirothipan]

related PR https://github.com/wso2/wso2-synapse/pull/1128

[Nirothipan]

This issue seems to be the consequence of https://github.com/wso2/product-ei/issues/2358

[abeykoon]

I tested the scenario with attached CAR application.

1. Setup a 2 node EI cluster
2. Deploy the CAR app to both nodes
3. Deactivate MP on the node1, which is running the task.
4. Shutdown node1
5. Node2 picked up the MP that is not deactivated and it is working.

[abeykoon]

Rename the CAR file and use.

testProjectCompositeApplication_1.0.0.zip

Conf of Node1

conf.zip

[abeykoon]

This issue happens when artifacts are added from management console

Change the setup as follows

1. Create a sim link to EI node01's deployment folder from node02's deployment folder.
2. when node01 is shut down, deactivated message processors are automatically getting activated at node02.