[
https://issues.apache.org/jira/browse/FELIX-4258?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13782028#comment-13782028
]
Holger Hoffstätte commented on FELIX-4258:
------------------------------------------
This sounds familiar :)
Would it help or be possible to hold classloaders in WeakReferences, which are
typically signalled very aggressively? Instead of releasing them normally and
hoping for the best (esp. with System.gc() possibly being disabled) this
(coupled with a separate reference queue that buries the dead) might do the
trick.
> Native library problems when the framework is quickly restarted
> ---------------------------------------------------------------
>
> Key: FELIX-4258
> URL: https://issues.apache.org/jira/browse/FELIX-4258
> Project: Felix
> Issue Type: Bug
> Affects Versions: framework-4.2.1
> Reporter: Guillaume Nodet
>
> If the framework is restarted quickly enough (using stop on the bundle 0 with
> a restart for example), native libraries are not unloaded and the bundle
> cache for those bundles is not refreshed leading to exceptions thrown when
> loading native libraries.
> {code}
> ERROR: Bundle org.apache.servicemix.bundles.snappy-java [113] Error starting
> mvn:org.apache.servicemix.bundles/org.apache.servicemix.bundles.snappy-java/1.0.4.1_1
> (org.osgi.framework.BundleException: Activator start error in bundle
> org.apache.servicemix.bundles.snappy-java [113].)
> java.lang.UnsatisfiedLinkError: Native Library
> /Users/sjavurek/Fuse/JBossFuse/jboss-a-mq-6.0.0.redhat-024/data/cache/bundle113/version0.0/bundle.jar-lib/0/org/xerial/snappy/native/Mac/x86_64/libsnappyjava.jnilib
> already loaded in another classloader
> at java.lang.ClassLoader.loadLibrary0(ClassLoader.java:1792)
> at java.lang.ClassLoader.loadLibrary(ClassLoader.java:1716)
> at java.lang.Runtime.loadLibrary0(Runtime.java:823)
> at java.lang.System.loadLibrary(System.java:1045)
> at
> org.xerial.snappy.SnappyBundleActivator.start(SnappyBundleActivator.java:33)
> at
> org.apache.felix.framework.util.SecureAction.startActivator(SecureAction.java:645)
> at org.apache.felix.framework.Felix.doActivateBundle(Felix.java:2317)
> at org.apache.felix.framework.Felix$7.call(Felix.java:2255)
> at org.apache.felix.framework.Felix$6.call(Felix.java:2199)
> at java.util.concurrent.FutureTask$Sync.innerRun(FutureTask.java:303)
> at java.util.concurrent.FutureTask.run(FutureTask.java:138)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:895)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:918)
> at java.lang.Thread.run(Thread.java:680)
> {code}
> Not sure what the best way to fix this, as performing a systematic refresh
> when the framework starts or stops may not be a good idea.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
