Tibco 无法追踪的断开连接

Question

In our production environment we are facing a strange problem with Tibco and Spring JMS. Randomly during the week our MessageListener implementation of Spring JMS (Spring Boot 2.5.2) no longer receives messages from the server although queues have messages. However, no client-side disconnection error is reported so the listener is still listening.

Do you have an idea? Does Spring JMS have a "auto reconnection" mechanism without explicit disconnection by the server? Or other idea to solve the problem? We also use com.tibco.tibjms tibjms 8.0.0 jar and tibco-ems tibcrypt 4.1 jar.

thank a lot, Adrien

Answer 1

Do you set the receiveTimeout on the org.apache.camel.component.jms.JmsComponent ? This timeout specifies how long the message listener will wait to receive a message. If this timeout expires, then the message listener will be dropped and reestablished.

The default value is one second, which is often much to low as this causes frequent connection drops and re-connections when there are no messages on the destination.

Answer 2

A possible explanation to this behavior is a loss of connectivity at.network level that would be undetected in the EMS client library.

To manage this you need to do two things:

#1 At EMS server level add the following properties in tibemsd.conf file:

client_heartbeat_server = 20

server_timeout_client_connection = 90

server_heartbeat_client = 20

client_timeout_server_connection = 90

Those properties enable some EMS mechanisms that allows to detect the loss of connectivity between the EMS server and the client applications.

#2 At your application level you should add the following properties in the JVM parameters:

-Dcom.tibco.tibjms.connect.attempts 6, 10000 -Dcom.tibco.tibjms.reconnect.attempts 12, 10000

Also if you are using connection factories you should add the following properties to each of the factories used by your application:

addprop factory <YourFactory> url=tcp://<YourServer1>:7222,tcp://<YourServer2>:7222

addprop factory <YourFactory> connect_attempt_count=6

addprop factory <YourFactory> connect_attempt_delay=10000

addprop factory <YourFactory> connect_attempt_timeout=1000

addprop factory <YourFactory> reconnect_attempt_count=12

addprop factory <YourFactory> reconnect_attempt_delay=10000

addprop factory <YourFactory> reconnect_attempt_timeout=1000

Those properties control the reconnection mechanisms available in the EMS client library.

With the above you should be able to detect loss of connectivity and reconnect automatically.