Aleksandar Kostadinov wrote: > Hallo, > > The real failure is build #1018 where it fails to start the > cluster-classloader-leak-0 server. > In both this run and the #1019 run which also failed in starting a server, there is some problem in starting the HAPartition service. Following is more for Bela's/Vladimir's info in case they see something. From #1018: 2008-10-23 21:49:12,153 DEBUG [org.jboss.ha.framework.server.JChannelFactory] (main) Passing unique node id 10.18.124.47:1099 to the channel as additional data 2008-10-23 21:49:12,172 DEBUG [org.jboss.cache.RPCManagerImpl] (main) Created Multiplexer Channel for cache cluster DefaultPartition-HAPartitionCache using stack udp 2008-10-23 21:49:12,183 INFO [STDOUT] (main) ------------------------------------------------------- GMS: address is 10.18.124.47:56595 ------------------------------------------------------- 2008-10-23 21:49:12,184 DEBUG [org.jboss.cache.RPCManagerImpl] (main) Block received at 10.18.124.47:56595 2008-10-23 21:49:12,184 DEBUG [org.jboss.cache.RPCManagerImpl] (main) Block processed at 10.18.124.47:56595 2008-10-23 21:49:14,278 INFO [org.jboss.cache.RPCManagerImpl] (main) Received new cluster view: [10.18.124.47:56595|0] [10.18.124.47:56595] 2008-10-23 21:49:14,279 DEBUG [org.jboss.cache.RPCManagerImpl] (main) UnBlock received at 10.18.124.47:56595 2008-10-23 21:49:14,279 DEBUG [org.jboss.cache.RPCManagerImpl] (main) UnBlock processed at 10.18.124.47:56595 2008-10-23 21:49:14,284 DEBUG [org.jboss.cache.RPCManagerImpl] (main) connected, state was retrieved successfully (in 2108 milliseconds) 2008-10-23 21:49:14,284 INFO [org.jboss.cache.RPCManagerImpl] (main) Cache local address is 10.18.124.47:56595 2008-10-23 21:49:14,289 DEBUG [org.jboss.cache.CacheStatus] (main) Ignoring call to start() as current state is INSTANTIATED 2008-10-23 21:49:14,289 DEBUG [org.jboss.cache.CacheStatus] (main) start() called while current state is INSTANTIATED -- call create() first 2008-10-23 21:49:14,289 DEBUG [org.jboss.cache.CacheStatus] (main) Ignoring call to create() as current state is STARTING 2008-10-23 21:49:14,291 DEBUG [org.jboss.cache.CacheStatus] (main) Ignoring call to start() as current state is STARTING 2008-10-23 21:49:14,291 DEBUG [org.jboss.cache.jmx.CacheJmxWrapper] (main) Registering interceptors 2008-10-23 21:49:14,309 INFO [org.jboss.cache.factories.ComponentRegistry] (main) JBoss Cache version: JBossCache 'Poblano' 2.2.1.GA 2008-10-23 21:52:44,169 INFO [org.jboss.bootstrap.microcontainer.ServerImpl] (JBoss Shutdown Hook) Runtime shutdown hook called, forceHalt: true From #1019 2008-10-24 09:21:25,881 INFO [STDOUT] (JBoss System Threads(1)-3) ------------------------------------------------------- GMS: address is 10.18.124.27:59288 ------------------------------------------------------- 2008-10-24 09:21:25,881 DEBUG [org.jboss.ha.framework.interfaces.HAPartition.DefaultPartition] (JBoss System Threads(1)-3) Block processed at null 2008-10-24 09:21:27,923 DEBUG [org.jboss.ha.framework.interfaces.HAPartition.DefaultPartition] (JBoss System Threads(1)-3) ViewAccepted: initial members set for partition DefaultPartition: 0 ([10.18.124.27:1099]) 2008-10-24 09:21:27,923 INFO [org.jboss.ha.framework.interfaces.HAPartition.DefaultPartition] (JBoss System Threads(1)-3) Number of cluster members: 1 2008-10-24 09:21:27,923 INFO [org.jboss.ha.framework.interfaces.HAPartition.DefaultPartition] (JBoss System Threads(1)-3) Other members: 0 2008-10-24 09:21:27,924 DEBUG [org.jboss.ha.framework.interfaces.HAPartition.DefaultPartition] (JBoss System Threads(1)-3) Unblock processed at null An important thing to note here is this code is actually concurrently starting two channels with the same stack name and shared transport. This will be the initial use of this shared transport. Note in the two log sets the thread actually doing the work is different; in the first its "main" connecting the channel used by the HA-JNDI/DistributedState JBC instance, in the latter its a system thread pool thread starting the channel HAPartition uses. In the #1019 logging does the "null" in "Block processed at null" ring any alarm bells? I'll keep poking. > #1017 issue is that a test name is output twice > http://hudson.qa.jboss.com/hudson/job/JBoss-AS-5.0.x-TestSuite-sun15/1017... > > [junit] Running > org.jboss.test.cluster.classloader.leak.test.FieldGranularityClassloaderLeakUnitTestCase > > [junit] Running > org.jboss.test.cluster.classloader.leak.test.FieldGranularityClassloaderLeakUnitTestCase > > [junit] Tests run: 1, Failures: 0, Errors: 1, Time elapsed: 0 sec > [junit] Test > org.jboss.test.cluster.classloader.leak.test.FieldGranularityClassloaderLeakUnitTestCase > FAILED (timeout) > > This is a junit/ant bug. Otherwise the build shows 14 tests fixed. > I'll try to put a workaround in the test count checks for that situation. > > Regards, > Aleksandar > _______________________________________________ > jboss-development mailing list > jboss-development(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/jboss-development

On Fri, Oct 24, 2008 at 5:02 PM, Brian Stansberry <brian.stansberry(a)redhat.com&gt; wrote: > As a temporary workaround to this I'm going to disable the concurrent > connection of these two channels. There's definitely something not right > going on in there. I wanted to point out that I did switch to the new server.xml and context.xml parsing for the web configuration. This doesn't seem to be the problem at all here, but you never know ...