home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cs.utexas.edu!ut-emx!ibmchs!auschs!awdprime.austin.ibm.com!konopik.austin.ibm.com!konopik
- From: konopik@konopik.austin.ibm.com (Brad Konopik)
- Newsgroups: comp.unix.aix
- Subject: Re: Question: What happens when a Terminal is powered-down?
- Keywords: aix sigterm runaway pseudoterminals dynix UniVerse powerdown
- Message-ID: <1992Nov19.174614.20323@awdprime.austin.ibm.com>
- Date: 19 Nov 92 17:46:14 GMT
- References: <319@westmx.westmont.edu>
- Sender: news@awdprime.austin.ibm.com (USENET News)
- Organization: IBM AIX Porting Center, Austin
- Lines: 43
-
- In article <319@westmx.westmont.edu> rodkey@westmont.EDU (John Rodkey) writes:
- >
- >I have this setup: RS/6000 mod 530, AIX 3.1.6, UniVerse 5.4, DYNIX v. 1??
- >Access to the RS/6000 from terminals is through Emulex and Lantronix
- >Network terminal servers. This problem occurs on both, so I have made
- >the assumption that the NTSs are not the weak link.
- >
- >The terminals (wyse 50's, PCs w/ emulation, etc.) are accessing the
- >RS/6000 in order to run the Library public access catalog, which is one
- >of the options in a ksh menu program.
- >DYNIX, the catalog program, is using the UniVerse implementation of
- >Pick, which is in turn layered over AIX.
- >
- >Here's the problem: when the users are done with the public catalog,
- >they sometimes turn off the terminal, which creates a 'cpu-hog' on our
- >system iff the user was using DYNIX at the time the plug was pulled.
- >
- >By CPU hog, I mean that its number of cpu clicks on ps -ef, C column is
- >between 20 and 120 (normally a cpu-intensive process shows about 12
- >clicks - but if there are 20 hogs, each may have only 9, for example).
- >
- >Clearly I need to get to the bottom of this problem. Killing processes
- >is repetitive, timewasting, etc.
- >
- >SO! The question is: what signals are issued to processes associated
- >with a pty when the terminal powers down? How does one handle those
- >signals to avoid the problem described? How does ksh handle these differently
- >than sh?
- >
-
- Rather than answer your question directly, I'll just tell you that Austin
- Defect Support's tty guy made a recent change in how orphaned processes
- are handled, to eliminate alot of the CPU hog stuff. Looking at the
- problem tracking information, I see that this fix is not yet built for
- AIX 3.1 (so it's going to be a long time before you see it). If you
- move to AIX 3.2, you'll want to ask for PTF U406391 (for APAR ix23574).
- Abstract: PROCESSES NOT KILLED WHEN SHELL, TELNET, TN, RLOGIN DIES.
-
- --
- tcpnet: konopik@konopik.austin.ibm.com | Brad Konopik
- ibmvnet: KONOPIK at AUSTIN | IBM AIX Porting Center, Austin
- internet: konopik.austin.ibm.com!konopik@ibmpa.awdpa.ibm.com
- uunet: ..!uunet!ibmsupt!ibmpa!konopik.austin.ibm.com!konopik
-