[SYNAPSE-846] Few parts of the code that prevents normal Synapse termination when running as a web application

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 2.1
• Fix Version/s: 3.0.0
• Environment: Synapse is acting as a web application inside web container (Tomcat), not tested as stand alone (but may be affected also).
• Assignee: Hiranya Kasub Jayathilaka
• Reporter: Lukas Valenta
• Created: 21/Feb/12 09:41
• Updated: 18/Aug/13 00:20
• Resolved: 15/Jul/13 21:03

Description

When using synapse as a web application, I noticed that it is unable to stop (stop web application) sucessfully. It leaves many threads running, which prevents web application and Tomcat itself from shutdown. I tracked those threads and found parts of code which "creates them but do not stop them". I created my own patch for those code parts and now application stops correctly. I would be pleased if those changes appears in some future version so I will have no need to hold "my own set of patches synapse and axis 2 classes". Changes to source codes of Synapse 2.1:

• org.apache.synapse.core.axis2.SynapseStartUpServlet:80 (after serverManager.stop():

serverManager.shutdown(); // this must be called also

• org.apache.synapse.transport.nhttp.HttpCoreNIOListener in method stop(), just before the end of try-catch block:

mbeanSupport.unregister();
metrics.destroy();

• org.apache.synapse.endpoints.AbstractEndpoint in method destroy() before "this.initialized = false":

if (metricsMBean != null)

{ metricsMBean.destroy(); metricsMBean = null; }

• org.apache.axis2.transport.jms.ServiceTaskManager (I know this belongs to Axis2, but it is the same problem):

• in the method stop() after the initial logging:

if (workerPool != null) {
try

{ workerPool.shutdown(1000); }

catch (InterruptedException e)

{ log.error("Stopping WorkerPool failed", e); }

}

• in the method stop() after "sharedConnection.stop();":

sharedConnection.close();

Thank you