[KARAF-4686] ClassLoader leak with RmiRegistryFactory and sun.rmi.transport.tcp.TCPEndpoint

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.4.4, 4.0.6, 3.0.8
• Fix Version/s: 4.1.0, 4.0.7
• Assignee: Guillaume Nodet
• Reporter: Grzegorz Grzybek
• Created: 31/Aug/16 15:47
• Updated: 12/Sep/16 06:58
• Resolved: 06/Sep/16 12:11

Description

Here's a leak I found in JVisualVM.

this     - value: org.apache.felix.framework.BundleWiringImpl #3
 <- m_wiring     - class: org.apache.felix.framework.BundleWiringImpl$BundleClassLoaderJava5, value: org.apache.felix.framework.BundleWiringImpl #3
  <- <classLoader>     - class: org.apache.karaf.management.RmiRegistryFactory$KarafServerSocketFactory, value: org.apache.felix.framework.BundleWiringImpl$BundleClassLoaderJava5 #3
   <- <class>     - class: org.apache.karaf.management.RmiRegistryFactory$KarafServerSocketFactory, value: org.apache.karaf.management.RmiRegistryFactory$KarafServerSocketFactory class RmiRegistryFactory$KarafServerSocketFactory
    <- ssf     - class: sun.rmi.transport.tcp.TCPEndpoint, value: org.apache.karaf.management.RmiRegistryFactory$KarafServerSocketFactory #1
     <- key     - class: java.util.HashMap$Entry, value: sun.rmi.transport.tcp.TCPEndpoint #9
      <- next     - class: java.util.HashMap$Entry, value: java.util.HashMap$Entry #19809
       <- [0]     - class: java.util.HashMap$Entry[], value: java.util.HashMap$Entry #28136
        <- table     - class: java.util.HashMap, value: java.util.HashMap$Entry[] #4031
         <- localEndpoints (sticky class)     - class: sun.rmi.transport.tcp.TCPEndpoint, value: java.util.HashMap #4065

sun.rmi.transport.tcp.TCPEndpoint has non-weak map that keeps references to socket factories: http://hg.openjdk.java.net/jdk8u/jdk8u/jdk/file/48c99b423839/src/share/classes/sun/rmi/transport/tcp/TCPEndpoint.java#l148

This prevents cleaning up classloader of org.apache.karaf.management.server bundle.

Issue Links

• links to: GitHub Pull Request #233
• links to: GitHub Pull Request #234