Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!utgpu!water!watmath!clyde!cbosgd!ucbvax!ucbcad!ames!hao!oddjob!mimsy!umd5!decuac!felix!zemon From: zemon@felix.UUCP Newsgroups: comp.unix.ultrix Subject: Exiting processes Message-ID: <11061@felix.UUCP> Date: Thu, 29-Oct-87 12:59:56 EST Article-I.D.: felix.11061 Posted: Thu Oct 29 12:59:56 1987 Date-Received: Sun, 1-Nov-87 00:55:50 EST Sender: zemon@felix.UUCP Organization: Ithaca College Lines: 62 Approved: zemon@felix.UUCP Reply-Path: Prelims: We are currently running a VAX 11/750 with Ultrix V2.0-1 Problem: processes. Occaisionally I will get a terminal that won't die because of an exiting process. Usually there is a user associated with that process/terminal, but in this example there isn't. What follows are the results from a "pstat -t" and "ps axut17" repectively. (the pstat command has been trimmed) # RAW CAN OUT MODE ADDR DEL COL STATE PGRP DISC 16 0 0 0 540500d8 8049e120 0 0 W 0 17 0 0 0 540540f0 8049e120 0 15 O H 22124 18 0 0 0 540500d8 8049e120 0 0 W 0 USER PID %CPU %MEM SZ RSS TT STAT TIME COMMAND root 22124 0.0 0.6 73 28 17 D 0:05 I am able to get rid of the process by using the new and improved file command. (Yes, I said "file" command.) If there is a user associated with the hung terminal, and they have mesg n, I must be superuser in order for this to work. When I type the following: %file /dev/tty17 The command will act as if it is in some kind of loop. After a second or two I use a Control-C to break out of it. When I look back at the terminal's status I now get the following. [The COL number, on the pstat output, doesn't seem to bother the terminals ability to log on to. That is, anyone can sign on to that port now.] # RAW CAN OUT MODE ADDR DEL COL STATE PGRP DISC 16 0 0 0 540500d8 8049e120 0 0 W 0 17 0 0 0 540540f0 8049e120 0 15 W 0 18 0 0 0 540500d8 8049e120 0 0 W 0 USER PID %CPU %MEM SZ RSS TT STAT TIME COMMAND root 25522 0.0 0.6 73 25 17 I 0:00 - T9600 tty17 (getty) If I take a look at lastcomm, it will show me what exactly the "file" command killed. When there is a specific user associated with the hung port the killed process is usually csh. cat shutter tty24 0.11 secs Wed Oct 28 08:23 getty S X root tty17 5.54 secs Tue Oct 27 15:59 file X shutter tty24 0.18 secs Wed Oct 28 08:23 ps shutter tty24 4.45 secs Wed Oct 28 08:22 I've talked to DEC about this and they're clueless. I'm hoping someone else out there has run into this problem and has either come up with a solution or at least knows why it is happening. If it helps at all, the ports this is happening on are dmf's with the following flags in the config file: device dmf1 at uba0 csr 0160400 flags 0x00 \ vector dmfsrint dmfsxint dmfdaint dmfdbint dmfrint dmfxint dmflint device dmf2 at uba0 csr 0160440 flags 0x00 \ vector dmfsrint dmfsxint dmfdaint dmfdbint dmfrint dmfxint dmflint Thank you for any light you are able to shed on this problem. Wendy Shutter (bitnet: shutter@ithaca) Ultrix System Manager Ithaca College