Hi, I debugged a bit, and it looks like it never calls the shutdown method in PooledTaskRunner after all. It gets stuck in DefaultMessageListenerContainer.doShutdown() where it calls this.lifecycleMonitor.wait(). Apparently, this does not terminate. I am not sure if this is a bug in Camel, ActiveMQ or Spring, or if it is even intended behaviour.
Anyway, setting the timeout to camel does the trick: camelContext.getShutdownStrategy().setTimeout(1L); Apparently the default for this is 5 minutes and I was never patient enough to wait so long. ;) Thanks for the help! Anna -----Original Message----- From: Ganesh Murthy [mailto:gmur...@redhat.com] Sent: Mittwoch, 30. September 2015 15:12 To: users@activemq.apache.org Subject: Re: Problem shutting down activemq client with failover transport How is Camel setting the shutdown flag on the PooledTaskRunner? Is Camel calling the shutdown() function to shut it down? The shutdown() method (which in turn calls shutdown(0)) on PooledTaskRunner is guaranteed to shut it down because if shutdown is set to true, then iterating is set to false. Have you tried the "timeout" parameter in FailoverTransport? http://activemq.apache.org/failover-transport-reference.html If you are using an IDE like IntelliJ, try putting in breakpoints on only that PooledTaskRunner thread and analyze what is going on. Thanks. -----Original Message----- From: Ganesh Murthy [mailto:gmur...@redhat.com] Sent: Dienstag, 29. September 2015 16:52 To: users@activemq.apache.org Subject: Re: Problem shutting down activemq client with failover transport Can you please provide a thread dump? That would throw some light on exactly what is going on. Thanks. ----- Original Message ----- From: "Anna Maier" <a.ma...@topdesk.com> To: users@activemq.apache.org Sent: Tuesday, September 29, 2015 9:31:13 AM Subject: Problem shutting down activemq client with failover transport Hi, I am using Camel to connect to ActiveMQ and it fails to shutdown when the ActiveMQComponent cannot make a connection to the broker with the failover protocol (activemq version 5.12.0). The problem seems to be that the failover transport tries to connect to the broker in an endless loop on its own thread that cannot be interrupted. I think the problem is in the PooledTaskRunner: there is a shutdown field which is set to true when camel shuts down, but it appears as false to the thread that is running the endless loop. I am currently at a loss what to do. I want to keep the maximum retries setting for the failover transport on indefinite, since I do not know when the the broker is back online. At the same time, it has to be possible to shut down the application even if there never was a connection to the ActiveMQ broker. Has anybody seen this problem before and can give some advice? Is this a bug in ActiveMQ? Attached there is some code that reproduces the problem - it never terminates. Regards, Anna The code: public class ShutdownBug { public static void main(String[] args) throws Exception { DefaultCamelContext camelContext = new DefaultCamelContext(); ActiveMQComponent activeMQComponent = new ActiveMQComponent(); activeMQComponent.setBrokerURL("failover:(http://localhost:61616)"); camelContext.addComponent("activemq", activeMQComponent); camelContext.addRoutes(new RouteBuilder() { @Override public void configure() { from("activemq:queue:myqueue").to("stream:out"); } }); Thread startThread = new Thread(new Runnable() { @Override public void run() { try { camelContext.start(); } catch (Exception e) { e.printStackTrace(); } } }); startThread.start(); startThread.interrupt(); startThread.join(); camelContext.stop(); } }
