You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@cxf.apache.org by "Freeman Yue Fang (Jira)" <ji...@apache.org> on 2020/04/16 16:02:00 UTC
[jira] [Commented] (CXF-8259) Blocked thread of JMS transport
during jboss shutdown
[ https://issues.apache.org/jira/browse/CXF-8259?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17085036#comment-17085036 ]
Freeman Yue Fang commented on CXF-8259:
---------------------------------------
I believe this is caused by the CXF-8161, especially this 5a4fd020bc87a552345425541bb62a35b0e5d8b7
{code}
- protected void restartConnection() {
+ protected synchronized void restartConnection() {
{code}
And restartConnection method will always run if something wrong during restartConnection(such as Jms broker is already down in the container) and never release the monitor lock of JMSDestination instance, since JMSDestination is subclass of org.apache.cxf.transport.AbstractObservable, so org.apache.cxf.transport.AbstractObservable will never get the moniter entry lock to finish the endpoint shutdown process. This is how the deadlock comes from.
Freeman
> Blocked thread of JMS transport during jboss shutdown
> -----------------------------------------------------
>
> Key: CXF-8259
> URL: https://issues.apache.org/jira/browse/CXF-8259
> Project: CXF
> Issue Type: Bug
> Components: JMS
> Affects Versions: 3.3.5, 3.3.6
> Reporter: Alessandro Proscia
> Priority: Major
> Attachments: jboss.log
>
>
> We have a soap over jms webservice implemented with a camel route (Camel 3.2.0) and CXF that uses activemq as message broker. The webservice is deployed on a jboss eap 7.1 with jdk 1.8. Following the CXF upgrade from version 3.3.4 to 3.3.5 (and also with 3.3.6), jboss hangs on shutdown due to a blocking thread of CXF:
> "ServerService Thread Pool – 82" #231 prio=5 os_prio=0 tid=0x0000000002981800 nid=0x66d2 waiting for monitor entry [0x00007f0d60d2f000]
> java.lang.Thread.State: BLOCKED (on object monitor)
> at org.apache.cxf.transport.AbstractObservable.setMessageObserver(AbstractObservable.java:42)
> Therefore, we can have to kill the Jboss process to complete the shutdown.
> The connection to activemq is managed by a resource adapter on jboss. Url is configured with a failover as follows:
> failover://(tcp://amqdev.internal:61616,tcp://amqdev.internal:61616,tcp://amqdev.internal:61616)?timeout=1000&initialReconnectDelay=2000&maxReconnectAttempts=2
> Attached you can find the thread dump taken from logs.
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)