4 Replies Latest reply on Nov 21, 2005 3:24 PM by Wilson Northrup

    killing Windows Processes

      I'm attempting to kill a hung process on a Windows server, but I'm not having any luck.


      felix2% kill -9 7068

      kill: kill 7068 failed: no error


      There is a Windows Resource Kit that contains a kill command, however this Resource Kit is not installed on every machine, and I'm told there are inconsistencies in its effectiveness.


      Any ideas how I can get rid of it?

        • 1. Re: killing Windows Processes

          Are you running the built-in kill that comes with nsh? If so, it should work if you cd //server and then kill the process.



          R51-DEMO-63% ps -ef | grep notepad
          joe 4872 0.0 0.0 732 3048 0 18:27:59 0:00 notepad
          R51-DEMO-63% kill 4872
          kill: kill 4872 failed: no such process
          R51-DEMO-63% cd //r51-demo-63
          r51-demo-63% ps -ef | grep notepad
          joe 4872 0.0 0.0 732 3048 0 18:27:59 0:00 notepad
          r51-demo-63% kill 4872


          • 2. Re: killing Windows Processes

            I just tried that and it didn't work at all with the built-in 'kill' command, with or without cd'ing to the localhost.



            • 3. Re: killing Windows Processes

              It works with some processes, but sometimes for strangely-owned processes, it won't work. They're the same processes for which an "End Process" in the Task Manager also fails. There's a "kill" command in the Windows Resource Kit. But since we don't have that installed on all servers, we were hoping that someone could suggest an alternative.

              • 4. Re: killing Windows Processes

                Using the BL kill usually works for me. Actually using the microsoft kill can be quite annoying (why do they change the name from "kill" to "taskkill" anyway? As if I didn't have enough problems).


                Anyway, my work around for those servers where i can not have extra utils deployed is to deploy insert required util here to somewhere like /tmp/blade/util, do what is required and what i need then do another job to wack the /tmp/blade/util directory when finished.


                if its only one server then i'll just cp the file from the respository where i keep all that stuff, and delete it when i'm done. Not elegant but it works. if you're not restricted to nsh, then ofcoure there are other ways native to the platform.