1. 08 7月, 2009 1 次提交
  2. 13 5月, 2009 2 次提交
  3. 07 2月, 2009 1 次提交
  4. 11 12月, 2008 1 次提交
  5. 02 10月, 2008 1 次提交
  6. 03 6月, 2008 1 次提交
  7. 24 5月, 2008 1 次提交
    • K
      [FIXED HUDSON-1458] Fixed the mysterious JVM hang issue on Windows&Maven2. · 704343f5
      kohsuke 提交于
      I spent a good part of yesterday afternoon to track down this problem. I first attached Visual Studio to the hanging JVM. While I can see the JVM source code and Windows API symbols all right, I couldn't figure out how to relate back the code produced by JIT to its Java, so this made the debugging very hard.
      
      What I observed was that Maven's main thread seems to be blocked in WaitForSingleObject kernel API, but I can't figure out what handle it's blocking on. This call appears to come directly from JIT-ed Java code.
      
      The problem disappears if I enable Java debugger support, or if I attach jconsole. So this cannot be a plain Java-level deadlock.
      
      So after that much work, I still coudldn't figure out the root cause.
      
      
      But I took the advice from mriem@dev.ajva.net anyway, and modified the remoting to use a separate TCP socket, instead of stdin/stdout. This is useful anyway, because sometimes JVM writes out to stdout and that breaks the whole set up, which was one of the known issues.
      
      This commit brings in this change, and in the limited testing so far, this appears to work. So let's hope that it fixes the problem once and for all.
      
      
      git-svn-id: https://hudson.dev.java.net/svn/hudson/trunk/hudson/main@9615 71c3de6d-444a-0410-be80-ed276b4c234a
      704343f5
  8. 25 4月, 2008 1 次提交
  9. 19 8月, 2007 3 次提交
  10. 03 6月, 2007 1 次提交
  11. 07 5月, 2007 3 次提交
  12. 13 4月, 2007 1 次提交
  13. 16 3月, 2007 1 次提交
  14. 01 3月, 2007 1 次提交