1 2 3 Previous Next 33 Replies Latest reply on Jun 11, 2008 8:50 AM by dimitris Go to original post
      • 15. Re: HDScanner exception
        anil.saldhana

         

         /*
         * Note: the semantics of FileURLConnection object is that the
         * results of the various URLConnection calls, such as
         * getContentType, getInputStream or getContentLength reflect
         * whatever was true when connect was called.
         */
         public void connect() throws IOException {
         if (!connected) {
         try {
         filename = file.toString();
         isDirectory = file.isDirectory();
         if (isDirectory) {
         files = (List) Arrays.asList(file.list());
         } else {
        
        


        Scott, the last line....... Every method in FileURLConnection apparently calls initializeHeaders() internally which always calls connect(). So maybe as a start, we need to see which file in question leads to this NPE????

        UPDATE: the file is a directory.

        • 16. Re: HDScanner exception
          anil.saldhana

          So basically the following call

           Arrays.asList(file.list())
          


          is failing inside Sun VM. For JBossVFS, the call is just getLastModified. So we need to figure out for which directory, this NPE is happening.

           ArrayList(E[] array) {
           if (array==null)
           throw new NullPointerException();
           a = array;
           }
          


          This is the NPE we are seeing.
          IMO, we need to see why file.list() is returning null?

          I also wonder why NPE is preferred to be thrown rather than IllegalArgumentException.

          • 17. Re: HDScanner exception
            emuckenhuber

            Hmm - i tried that at home again and maybe it's just for 64Bit machines?
            I used an older Linux machine at home and it does not seem to be a problem there...

            • 18. Re: HDScanner exception
              anil.saldhana

              The dude that is causing the NPE is this:

              sun.net.www.protocol.file.FileURLConnection:file:/home/anil/jboss-5.0/jboss-head/build/output/jboss-5.0.0.CR1/server/default/deploy/management/console-mgr.sar/
              


              • 19. Re: HDScanner exception
                anil.saldhana

                After that, this one:

                FileHandler@781346814[path=jbossws.sar context=file:/home/anil/jboss-5.0/jboss-head/build/output/jboss-5.0.0.CR1/server/default/deploy/ real=file:/home/anil/jboss-5.0/jboss-head/build/output/jboss-5.0.0.CR1/server/default/deploy/jbossws.sar/]


                • 20. Re: HDScanner exception
                  anil.saldhana

                  Somehow once AS5 starts, the VFS HDScanner is continuously checking on the updates for console.sar and jbossws.sar and after some time (less than a minute), these directories come out with null files. Not sure what is happening.

                  Update: I see that the console.sar just failed once. After that there was no query for it. But the jbossws.sar is being continuously scanned for contents by the vfs layer.

                  • 21. Re: HDScanner exception
                    dimitris

                    Maybe that's random which directories will cause the NPE, at a point where it starts running out of file descriptors.

                    • 22. Re: HDScanner exception
                      dimitris
                      • 23. Re: HDScanner exception
                        starksm64

                        After 6 hours I'm not seeing this in the all config on osx either. Try reverting to the jboss-common-core 2.2.5.GA to see if that is the cause.

                        • 24. Re: HDScanner exception
                          anil.saldhana

                          Let me validate the roll back in my local workspace.

                          • 25. Re: HDScanner exception
                            anil.saldhana
                            • 26. Re: HDScanner exception
                              emuckenhuber

                               

                              "anil.saldhana@jboss.com" wrote:
                              Rollback works.

                              Although i applied the rollback it does not work for me - still getting the same failures :-/
                              Can you may try that again ?

                              • 27. Re: HDScanner exception
                                dimitris

                                It must have to do with maven still using the newer version. Nuke your thirdparty/jboss/common-core and rebuild. Make sure it uses the correct version.

                                • 28. Re: HDScanner exception
                                  emuckenhuber

                                   

                                  "dimitris@jboss.org" wrote:
                                  It must have to do with maven still using the newer version. Nuke your thirdparty/jboss/common-core and rebuild. Make sure it uses the correct version.


                                  right, i was missing that - works now. Thanks

                                  • 29. Re: HDScanner exception
                                    starksm64

                                    I have reproduced it on a new f9 install with the latest sun jdk:

                                    06:56:05,036 INFO [ServerInfo] Java version: 1.5.0_15,Sun Microsystems Inc.
                                    06:56:05,045 INFO [ServerInfo] Java VM: Java HotSpot(TM) 64-Bit Server VM 1.5.0_15-b04,Sun Microsystems Inc.
                                    06:56:05,045 INFO [ServerInfo] OS-System: Linux 2.6.25-14.fc9.x86_64,amd64
                                    


                                    each hd scan is leaking 27 descriptors it seems, and the lsof count for the process just keeps growing.
                                    [starksm@banshee testsuite]$ lsof -p 20881 | wc
                                     146 1306 19918
                                    [starksm@banshee testsuite]$ lsof -p 20881 | wc
                                     128 1142 17046
                                    [starksm@banshee testsuite]$ lsof -p 20881 | wc
                                     151 1349 21177
                                    [starksm@banshee testsuite]$ lsof -p 20881 | wc
                                     178 1592 25874
                                    [starksm@banshee testsuite]$ lsof -p 20881 | wc
                                     205 1835 30571
                                    ...
                                    [starksm@banshee testsuite]$ lsof -p 20881 | wc
                                     259 2321 39965
                                    ...
                                    [starksm@banshee testsuite]$ lsof -p 20881 | wc
                                     529 4751 86935
                                    ...
                                    [starksm@banshee testsuite]$ lsof -p 20881 | wc
                                     1042 9368 176178
                                    


                                    I also see the same problem with the open jdk 1.6.0-b09 bundled with f9.
                                    I guess http://jira.jboss.com/jira/browse/JBCOMMON-55 should be reopened.