using JRockit R28.1.3-11-141760-1.6.0_24-20110301-1432-linux-x86_64 we experience a memory leak in our j2ee application eventually causing the application to stall. So far I only got to obtain 2 hprof dumps, which I analyzed with eclipse MAT. The MAT result is pretty clear but I'm not sure what the cause is. There's one object of com.sun.xml.ws.server.WSEndpointImpl with a growing retained heap, at the moment of the memory dump it was 929 MB (on the first dump it was ~300 MB). Here's the import part of the dominator tree:
Class Name | Shallow Heap | Retained Heap | Percentage
com.sun.xml.ws.server.WSEndpointImpl @ 0xafaf0f70 | 176 | 929.780.784 | 69,61%
|- com.sun.xml.ws.binding.SOAPBindingImpl @ 0xafaf1160 | 72 | 929.046.360 | 69,56%
| |- com.sun.xml.ws.binding.WebServiceFeatureList @ 0xafaf13e0 | 56 | 929.044.320 | 69,56%
| | '- java.util.HashMap @ 0xafaf1e58 | 80 | 929.044.264 | 69,56%
| | |- java.util.HashMap$Entry[16] @ 0xafaf3058 | 152 | 929.044.160 | 69,56%
| | | |- java.util.HashMap$Entry @ 0xaf9b8f48 | 48 | 928.998.912 | 69,55%
| | | | '- weblogic.wsee.jaxws.EndpointCreationInterceptorFeature @ 0xaf9b8f80| 32 | 928.998.864 | 69,55%
| | | | '- java.util.HashSet @ 0xaf9b92f8 | 24 | 928.998.832 | 69,55%
| | | | '- java.util.HashMap @ 0xaf9b9528 | 80 | 928.998.808 | 69,55%
| | | | |- java.util.HashMap$Entry[32768] @ 0x9462cb88 | 262.168 | 928.998.704 | 69,55%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0x9b672ae8 | 40 | 152.104 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0xad2a9360 | 40 | 152.104 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0xa3949490 | 40 | 152.104 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0xa3a332c8 | 40 | 152.104 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0x954ef698 | 40 | 151.440 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0xb3acff18 | 40 | 148.256 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0xa1db8a38 | 40 | 148.000 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0xb9d18ee8 | 40 | 148.000 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0xbe6bf4d0 | 40 | 147.896 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0xa0397080 | 40 | 147.896 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0xbf6693f0 | 40 | 147.896 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0xc3c4ddb0 | 40 | 143.792 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0x82714140 | 40 | 143.792 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0xa0347178 | 40 | 143.792 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0x919c9948 | 40 | 143.688 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0xbb1f6d28 | 40 | 135.960 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0x8c3cd118 | 40 | 135.488 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0x91d6e950 | 40 | 134.936 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0x8abefe80 | 40 | 134.936 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0x89f0aaa8 | 40 | 134.736 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0xb0f116a8 | 40 | 134.600 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0x9da69938 | 40 | 134.392 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0x96cae018 | 40 | 134.392 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0x88691950 | 40 | 134.040 | 0,01%
| | | | | |- weblogic.wsee.jaxws.ServerLateInitTube @ 0xc201f268 | 40 | 134.040 | 0,01%
'- Total: 25 of 36.803 entries
I would appreciate any hints on how to proceed as these are standard java / weblogic classes. If I drill down into the objects I see some ErrorHandler objects, so these objects might accumulate with erroneus requests, but I'm not positive on this since I only checked a few instances. Before I start to try to reproduce this on a test environment I would like to ask for your help.
Kind regards, Thomas
This is a known bug (ID 1303769.1), patch is available.