fredrick@acd.acd.ucar.edu (Tim Fredrick) (03/12/91)
I hope this is something simple. But when a terminal sessions is terminated abnormally (power down, telnet disconnect, etc.) many processes stay in the process tables. This is also true when you press CTRL-SHIFT-BACKSPACE in AIXWindows -- some X processes will stay in the process table as seen by "ps -efgaux". Some processes (like the game nethack) really need to get a "kill -HUP" signal during these conditions so they can do error recovery. Has anyone else experienced this? Is this a bug or a procedural problem? Any insight would be tremendously appreciated. Thanks. --Tim
bengsig@dk.oracle.com (Bjorn Engsig) (03/13/91)
Article <10571@ncar.ucar.edu> by fredrick@acd.acd.ucar.edu (Tim Fredrick) says: | |I hope this is something simple. But when a terminal sessions is terminated |abnormally (power down, telnet disconnect, etc.) many processes stay in |the process tables. Did you have a look at stty settings, in particular 'stty hupcl -clocal' could be useful (hangup on last close, and not connected local). -- Bjorn Engsig, ORACLE Corporation, E-mail: bengsig@oracle.com, bengsig@oracle.nl
mcguire@math.uiowa.edu (Charlie McGuire) (03/13/91)
In article <10571@ncar.ucar.edu>, fredrick@acd.acd.ucar.edu (Tim Fredrick) writes: |> |> I hope this is something simple. But when a terminal sessions is terminated |> abnormally (power down, telnet disconnect, etc.) many processes stay in |> the process tables. This is also true when you press CTRL-SHIFT-BACKSPACE |> in AIXWindows -- some X processes will stay in the process table as seen |> by "ps -efgau ..... Dealing with killing these things is a daily effort for me, and a real pain in the neck. I have no idea why it behaves this way. At the moment I have enough to do without IBM putting it on my back to demonstrate how to reproduce it. (Which is why I haven't called it in.) As often as it happens here, I find it hard to believe someone at IBM hasn't noticed it. *************************************************************************** * Charlie McGuire | INTERNET: mcguire@math.uiowa.edu * * Systems Programmer | mcguire@cs.uiowa.edu * * Computer Science Dept. | * * The University of Iowa | PHONE: (319) 335-2730 * ***************************************************************************
root@acd.uucp (03/14/91)
Tim Fredrick (me!) writes: >|> >|> I hope this is something simple. But when a terminal sessions is terminated >|> abnormally (power down, telnet disconnect, etc.) many processes stay in >|> the process tables. This is also true when you press CTRL-SHIFT-BACKSPACE >|> in AIXWindows -- some X processes will stay in the process table as seen >|> by "ps -efgau ..... Some have speculated to me that this is a problem with /bin/csh. One person wrote to say that the problem has been fixed with AIX3.1.5 which was announced and scheduled to ship March 22. Until then I guess we get to keep killing processes by hand -- I haven't tried to run processes under any of the other shells to see if we have a similar problem. Thanks. --Tim
jjs@heart.austin.ibm.com (Jim Shaffer) (03/15/91)
In article <10602@ncar.ucar.edu> root@acd.uucp writes: >Some have speculated to me that this is a problem with /bin/csh. One person >wrote to say that the problem has been fixed with AIX3.1.5 which was >announced and scheduled to ship March 22. Until then I guess we get to >keep killing processes by hand -- I haven't tried to run processes under >any of the other shells to see if we have a similar problem. Thanks. --Tim As far as I know this is a csh problem only. The problem with csh was that it didn't kill foreground children when it received a SIGHUP. It seems to be fixed in 3.1.5. Jim Shaffer cs.utexas.edu!ibmchs!auschs!morse.austin.ibm.com!jjs "Don't tell me I'm burning the candle at both ends -- tell me where to get more wax!!"