[
https://issues.apache.org/jira/browse/ARTEMIS-5873?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Timothy A. Bish resolved ARTEMIS-5873.
--------------------------------------
Fix Version/s: 2.52.0
Resolution: Fixed
> Potential infinite loop on shutdown
> -----------------------------------
>
> Key: ARTEMIS-5873
> URL: https://issues.apache.org/jira/browse/ARTEMIS-5873
> Project: Artemis
> Issue Type: Bug
> Components: Broker
> Affects Versions: 2.50.0
> Reporter: Emmanuel Hugonnet
> Priority: Major
> Labels: pull-request-available
> Fix For: 2.52.0
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
> When shutting down the broker the thread pool that holds scheduled tasks is
> stopped using
> [ExecutorService.shutdownNow()|https://docs.oracle.com/en/java/javase/17/docs/api/java.base/java/util/concurrent/ExecutorService.html#shutdownNow()].
> This method will _interrupt_ threads currently executing.
> If
> {{org.apache.activemq.artemis.core.server.impl.FileLockNodeManager.MonitorLock}}
> is running and if it is calling
> {{org.apache.activemq.artemis.core.server.impl.FileLockNodeManager#getState}}
> which, in turn, is calling
> {{java.nio.channels.FileChannel#read(java.nio.ByteBuffer, long)}} then this
> will result in a {{java.nio.channels.ClosedByInterruptException}} and the
> {{java.nio.channels.FileChannel}} _will be closed_. A stack-trace like the
> following will be logged:
> {noformat}
> ERROR [org.apache.activemq.artemis.core.server.impl.FileLockNodeManager]
> (Thread-2 (activemq-scheduled-default))
> java.nio.channels.ClosedByInterruptException:
> org.apache.activemq.artemis.core.server.NodeManager$NodeManagerException:
> java.nio.channels.ClosedByInterruptException
> at
> [email protected]//org.apache.activemq.artemis.core.server.impl.FileLockNodeManager.getState(FileLockNodeManager.java:375)
> at
> [email protected]//org.apache.activemq.artemis.core.server.impl.FileLockNodeManager$MonitorLock.run(FileLockNodeManager.java:539)
> at
> [email protected]//org.apache.activemq.artemis.core.server.ActiveMQScheduledComponent.runForExecutor(ActiveMQScheduledComponent.java:327)
> at
> [email protected]//org.apache.activemq.artemis.core.server.ActiveMQScheduledComponent.bookedRunForScheduler(ActiveMQScheduledComponent.java:342)
> at
> [email protected]//org.apache.activemq.artemis.core.server.ActiveMQScheduledComponent.runForScheduler(ActiveMQScheduledComponent.java:353)
> at
> [email protected]//org.apache.activemq.artemis.core.server.ActiveMQScheduledComponent.lambda$start$0(ActiveMQScheduledComponent.java:180)
> at
> java.base/java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:539)
> at
> java.base/java.util.concurrent.FutureTask.runAndReset(FutureTask.java:305)
> at
> java.base/java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:305)
> at
> java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1136)
> at
> java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:635)
> at
> [email protected]//org.apache.activemq.artemis.utils.ActiveMQThreadFactory$1.run(ActiveMQThreadFactory.java:120)
> Caused by: java.nio.channels.ClosedByInterruptException
> at
> java.base/java.nio.channels.spi.AbstractInterruptibleChannel.end(AbstractInterruptibleChannel.java:199)
> at
> java.base/sun.nio.ch.FileChannelImpl.endBlocking(FileChannelImpl.java:171)
> at
> java.base/sun.nio.ch.FileChannelImpl.readInternal(FileChannelImpl.java:844)
> at java.base/sun.nio.ch.FileChannelImpl.read(FileChannelImpl.java:821)
> at
> [email protected]//org.apache.activemq.artemis.core.server.impl.FileLockNodeManager.getState(FileLockNodeManager.java:360)
> ... 11 more{noformat}
> As the broker continues shutting down it will try to use this {{FileChannel}}
> again to set the status. However, using the aforementioned {{FileChannel}}
> will result in a {{java.nio.channels.ClosedChannelException}}. This exception
> will be logged, but it will not be thrown. Instead,
> {{org.apache.activemq.artemis.core.server.impl.FileLockNodeManager#lock}}
> will loop forever by default due to {{lockAcquisitionTimeoutNanos}} being set
> to {{-1}}. Here's an example stack trace logged during this process:
> {noformat}
> WARN [org.apache.activemq.artemis.core.server.impl.FileLockNodeManager]
> (ServerService Thread Pool -- 85) Failure when accessing a lock file:
> java.nio.channels.ClosedChannelException
> at
> java.base/sun.nio.ch.FileChannelImpl.ensureOpen(FileChannelImpl.java:159)
> at
> java.base/sun.nio.ch.FileChannelImpl.tryLock(FileChannelImpl.java:1314)
> at
> java.base/java.nio.channels.FileChannel.tryLock(FileChannel.java:1191)
> at
> [email protected]//org.apache.activemq.artemis.core.server.impl.FileLockNodeManager.tryLock(FileLockNodeManager.java:401)
> at
> [email protected]//org.apache.activemq.artemis.core.server.impl.FileLockNodeManager.lock(FileLockNodeManager.java:421)
> at
> [email protected]//org.apache.activemq.artemis.core.server.impl.FileLockNodeManager.writeFileLockStatus(FileLockNodeManager.java:336)
> at
> [email protected]//org.apache.activemq.artemis.core.server.impl.FileLockNodeManager.setPaused(FileLockNodeManager.java:318)
> at
> [email protected]//org.apache.activemq.artemis.core.server.impl.FileLockNodeManager.pausePrimaryServer(FileLockNodeManager.java:278)
> at
> [email protected]//org.apache.activemq.artemis.core.server.impl.PrimaryOnlyActivation.close(PrimaryOnlyActivation.java:106)
> at
> [email protected]//org.apache.activemq.artemis.core.server.impl.ActiveMQServerImpl.stop(ActiveMQServerImpl.java:1488)
> at
> [email protected]//org.apache.activemq.artemis.core.server.impl.ActiveMQServerImpl.stop(ActiveMQServerImpl.java:1245)
> at
> [email protected]//org.apache.activemq.artemis.core.server.impl.ActiveMQServerImpl.stop(ActiveMQServerImpl.java:1229)
> at
> [email protected]//org.apache.activemq.artemis.core.server.impl.ActiveMQServerImpl.stop(ActiveMQServerImpl.java:1062)
> at
> [email protected]//org.apache.activemq.artemis.core.server.impl.ActiveMQServerImpl.stop(ActiveMQServerImpl.java:1056)
> ...{noformat}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
