本人遭遇到该问题三次
9.2MP3
10.3
10.0MP1
On the IBM AIX platform, all the servers hang. The hang is not related to load, and usually occurs within 10 minutes or so of server startup. A thread dump reveals two deadlocked threads, one doing a socket close() and one doing a socket poll(). For example:
at java/net/PlainSocketImpl.socketClose0(Native Method)
at java/net/PlainSocketImpl.socketPreClose(PlainSocketImpl.java:744(Compiled Code))
at java/net/PlainSocketImpl.close(PlainSocketImpl.java:568(Compiled Code))
at java/net/SocksSocketImpl.close(SocksSocketImpl.java:1049(Compiled Code))
at java/net/Socket.close(Socket.java:1341(Compiled Code))
at weblogic/socket/WeblogicSocket.close(WeblogicSocket.java:80(Compiled Code))
at weblogic/socket/SocketMuxer.closeSocket(SocketMuxer.java:449(Compiled Code))
at weblogic/socket/SocketMuxer.cancelIo(SocketMuxer.java:773)
at weblogic/socket/SocketMuxer$TimerListenerImpl.timerExpired(SocketMuxer.java:976(Compiled Code))
at weblogic/timers/internal/TimerImpl.run(TimerImpl.java:273(Compiled Code))
at weblogic/work/SelfTuningWorkManagerImpl$WorkAdapterImpl.run(SelfTuningWorkManagerImpl.java:464(Compiled Code))
at weblogic/work/ExecuteThread.execute(ExecuteThread.java:197(Compiled Code))
at weblogic/work/ExecuteThread.run(ExecuteThread.java:172(Compiled Code))
'ExecuteThread: '7' for queue: 'weblogic.socket.Muxer'' (TID:0x3511B500, sys_thread_t:0x34FAD6C8, state:R, native ID:0x003E80CB) prio=5
at weblogic/socket/PosixSocketMuxer.poll(Native Method)
at weblogic/socket/PosixSocketMuxer.processSockets(PosixSocketMuxer.java:102)
at weblogic/socket/SocketReaderRequest.run(SocketReaderRequest.java:29)
at weblogic/socket/SocketReaderRequest.execute(SocketReaderRequest.java:42)
at weblogic/kernel/ExecuteThread.execute(ExecuteThread.java:145)
at weblogic/kernel/ExecuteThread.run(ExecuteThread.java:117)
Internally, WebLogic Server polls the sockets to see if there is any new activity. The trouble is we are also calling socket.close() to try and clear the sockets of existing clients at the same time.
According to IBM engineering:
- The underlying TCP implementation of socket close was changed between IBM JDK SR5 and SR7. Part of that change is that dup2 is now used.
- Because of the dup2 change, the poll() and close() methods cannot be called at the same time on the same socket.
This causes the deadlock.
Patches for WLS 8.x can be found in My Oracle Support. Open the Patches & Updates tab. Search for patch ID 8173442 for the patches for WLS 8.1mp3, 8.1mp4, and 8.1mp5. Search for patch ID 8179792 for the patch for WLS 8.1mp6.
Patches for WLS 9.x and higher can be downloaded from Smart Update using these patch IDs and passcodes:
------------------------------------------
PATCH REPOSITORY INFORMATION
------------------------------------------
WLS Version | Patch ID | Passcode
------------+----------+----------------
9.2 | T4DV | 7C7PYV9B
9.2mp1 | HZHQ | PTUYCCSI
9.2mp2 | WJD2 | GU1CW2AB
9.2mp3 | GNLT | 8J9L6Q4Y
10.0 | PMAJ | 9UQ69LLT
10.0mp1 | ITVL | K8RBHQQ2
10.3 | 9YT5 | I1DB5QSV
Fixed in: 10.0.2, 10.3.1
来自 “ ITPUB博客 ” ，链接：https://blog.itpub.net/96639/viewspace-662578/ ，如需转载，请注明出处，否则将追究法律责任。
