Sponsored Content
Top Forums UNIX for Advanced & Expert Users When kill doesnt work, how to kill a process ? Post 302073218 by Perderabo on Thursday 11th of May 2006 04:36:47 AM
Old 05-11-2006
If it has died and is now a zombie, killing the parent process will cause it to be re-parented to init which will reap it. But if it is sleeping at a high kernel priority, the only general answer is to reboot the system. Since you're on HP-UX, you should use glance to examine the process and determine what it is waiting for. Then there is some chance that you can correct it. For example, if it is trying to close a tape drive that has been powered off, bringing the tape drive back online may allow the close to succeed. Or maybe it is trying to write to /dev/console and someone typed an X-off (control s), so typing an X-on (control q) may allow the write to finish. If it is doing disk i/o to an NFS mounted filesystem with nointr, fixing the NFS server or the network may do it. Or maybe it is a bug in the kernel and so you want to identify what resource it is waiting for and then check for patches that look applicable. This won't help kill the particular process you have now, but patching the kernel would insure it doesn't keep on happening.
 

10 More Discussions You Might Find Interesting

1. Shell Programming and Scripting

how to start a process and make it sleep for 5 mins and then kill that process

how to start a process and make it sleep for 5 mins and then kill that process (6 Replies)
Discussion started by: shrao
6 Replies

2. Programming

kill(0,-9) don't kill the process

Hi all i have simple c program , when i wish to kill the app im using kill(0,-9) , but it seams this command don't do any thing and the program. just ignore it . what im doing wrong here ? im using HP-UX ia64 Thanks (9 Replies)
Discussion started by: umen
9 Replies

3. Solaris

How to Kill a Process

Hi, how can i kill one running process, how can i get the process id thanks .BJ (5 Replies)
Discussion started by: bennichan
5 Replies

4. Shell Programming and Scripting

When kill [pid] does not work...

Hi, On my Linux machine, using Bash, I sometimes run into a situation where doing the following does not seem to work at all. kermit@fastbox ~ $ ps -A | grep firefox-bin 5375 ? 00:06:57 firefox-bin <defunct> 5624 ? 00:00:00 firefox-bin kermit@fastbox ~ $ kill 5624... (7 Replies)
Discussion started by: kermit
7 Replies

5. Shell Programming and Scripting

Kill a process without using kill command

Sorry, posted the question in other forum. (0 Replies)
Discussion started by: sudhamacs
0 Replies

6. Linux

Kill a process without using kill command

I want to Kill a process without using kill command as i don't have privileges to kill the process. I know the pid and i am using Linux 2.6.9 OS. (6 Replies)
Discussion started by: sudhamacs
6 Replies

7. Shell Programming and Scripting

Shell Script to Kill Process(number of process) Unix/Solaris

Hi Experts, we do have a shell script for Unix Solaris, which will kill all the process manullay, it used to work in my previous env, but now it is throwing this error.. could some one please help me to resolve it This is how we execute the script (and this is the requirement) ... (2 Replies)
Discussion started by: jonnyvic
2 Replies

8. UNIX for Dummies Questions & Answers

Script to start background process and then kill process

What I need to learn is how to use a script that launches background processes, and then kills those processes as needed. The script successfully launches the script. But how do I check to see if the job exists before I kill it? I know my problem is mostly failure to understand parameter... (4 Replies)
Discussion started by: holocene
4 Replies

9. Solaris

Cannot kill a process with kill -9

Hello everyone, I have a process that I want to kill. I have tried kill-9 PID but it doesn't work. I have tried preap PID but it doesn't work too. The parent of my process is the process whose PID is 1, so I can't kill it. My OS is a Solaris 9. Can anyone help me understand what's going... (3 Replies)
Discussion started by: adilyos
3 Replies

10. Shell Programming and Scripting

Kill an specific process ID using the KILL and GREP commands

Good afternoon I need to KILL a process in a single command sentence, for example: kill -9 `ps -aef | grep 'CAL255.4ge' | grep -v grep | awk '{print $2}'` That sentence Kills the process ID corresponding to the program CAL255.4ge. However it is possible that the same program... (6 Replies)
Discussion started by: enriquegm82
6 Replies
BOOT(8) 						      System Manager's Manual							   BOOT(8)

NAME
boot - 2.11BSD bootstrap proceedure DESCRIPTION
The 2.11BSD system is started by a two-stage process. The first is a primary bootstrap (limited to 512 bytes) which is able to read in relatively small stand-alone programs; the second (called boot) is used to read in the system itself. The primary bootstrap must reside in block zero of the boot device (the disklabel resides in block one). It can be read in and started by standard ROM cold boot routines or, if necessary, by keying in a small startup routine. The primary bootstrap is capable of loading only type 0407 executable files (impure (non-shared), non-separate I&D.) Copies of the block zero bootstraps are kept in the directory /mdec. Disklabel(8) is normally used to place a copy of the appropriate bootstrap in block zero of new file systems. The primary bootstrap loads boot from the file system that starts at block 0 of the drive specified to the boot ROM. Normally the boot device is automatically used as the root filesystem. This action can be overriden by specifying the -R command to boot. If boot is not found the system will hang as the primary boot spins in an endless loop trying to find boot. No diagnostic message results if the file cannot be found. o In an emergency, the bootstrap methods described in the paper Installing and Operating 2.11BSD can be used to boot from a distribu- tion tape. The secondary boot program, called boot, actually brings in the system. When read into location 0 and executed, boot sets up memory man- agement, relocates itself into high memory, and types its name and a `:' on the console. If this is an automatic, unattended reboot, boot will use a default file specification for the installation, typing the file's name after the prompt. Otherwise, it reads a file specifica- tion from the console. Normal line editing characters can be used to make corrections while typing this (see below for file specification format). If only a carriage return is typed, a default name (/unix) will be used. Boot finds the [specified] file and loads it into mem- ory location zero, sets up memory management as required, and calls the program by executing a `trap' instruction. For the system to boot, /etc/init must exist and be executable; if it is not, the kernel will print a message to that effect and loop. Further, for a single user boot, the files /bin/sh and /dev/console must also exist and /bin/sh must be executable (if either of these is missing, init will attempt multi-user operation). For a multi-user boot the file /etc/ttys must exist (if missing, init will attempt sin- gle user operation). Init runs the autoconfig(8) program to probe for and initialize devices. Autoconfig only knows to look in /unix, thus if an alternate ker- nel name was specified none of the devices except /dev/console and the boot disk will be known. If autoconfig problems are suspected (or if you are simply voyeuristic) the debug flag can be turned on by specifying -D to boot (see below). When the system is running in single user mode, it starts a single user shell on the console which types a `#' prompt. After doing any file system checks and setting the date (date(1)) a multi-user system can be brought up by typing an EOT (control-d) in response to the `#' prompt. Boot file specification format: The file specifications used with boot are of the form: device(ctlr,unit,part)path [-aRrDs] or -bootcommand where device is the type of the device to be searched; ctlr is the controller number of the disk unit is the unit number of the disk or tape; part is the partition number of a filesystem on the specified disk or the tape file number if the device is a tape. The underlying device driver must support disklabels and a valid disklabel must be present if part is anything except 0. path is the path name of a disk file to be loaded with all mount prefixes stripped off (path must be omitted for tape files.) Tape files are separated by single tape marks. Flags to boot may be specified in either of two places. At the : prompt and after the file name. The options are: -a Ask for a kernel name. This is present for symmetry only because in order to specify this option you already have to be at the : prompt. -D Turn on the autoconfig debug flag. -R force the kernel to use its compiled in root device rather than adapting to the boot device. -s tell init to enter single user state rather than bringing the system all the way up to multi-user mode. -r mount the root filesys- tem read-only. This is not currently supported by the kernel mostly because pipes are implemented in the filesystem. Commands (-bootcommand) to boot are: -bootflags N where N is a decimal number. -bootflags flag where flag is from the list above. -bootdebug N where N is a decimal number. This is a general purpose flag word used by boot and is not passed to the loaded program or kernel. Device is one of the following xp RM02/03/05, RP04/05/06, DIVA, SI Eagle, CDC 9766, Fuji 160 rp RP03 rk RK05 hk RK06/7 rl RL01/2 si RM05, CDC 9766 ra RA60/80/81, RX50, RD51/52/53, RC25 ht TU/TE16 tm TU/TE10 ts TS-11 The stand alone tape drive unit number is specially encoded to specify both unit number and tape density (BPI). Most tape subsystems either automatically adjust to tape density or have switches on the drives to force the density to a particular setting, but for those which don't the following density select mechanisms may be necessary. The ts only operates at 1600BPI, so there is no special unit density encoding. The ht will operate at either 800BPI or 1600BPI. Units 0 through 3 correspond to 800BPI, and 4 through 7 to 1600BPI on drives 0 through 3 respectively. The standard DEC tm only supports 800BPI (and hence can't be used with the standard distribution tape), but sev- eral widely used tm emulators support 1600BPI and even 6250BPI. Units 0 through 3 correspond to 800BPI, 4 through 7 to 1600BPI, and 8 through 11 to 6250BPI on drives 0 through 3 respectively. For example, to boot a system from unit 0 on an RK07, type "hk(0,0)unix" to the boot prompt. The specification "ra(1,0)unix" indicates an MSCP disk, unit 1. The specification "ra(1,0,0)unix" indicates an MSCP disk, unit 0 but on controller 1. And finally the specification "ts(0,3)" would cause the fourth file on a tape threaded on `ts' tape drive 0 to be loaded and executed. Cold boot loaders: The following programs to load and execute the primary bootstrap may be installed in read-only memories or manually keyed into main memory. Each program is position-independent but should be placed well above location 0 so it will not be overwritten. Each reads a block from the beginning of a device into core location zero. The octal words constituting the program are listed on the left. RK (drive 0): 012700 mov $rkda,r0 177412 005040 clr -(r0) / rkda cleared by start 010040 mov r0,-(r0) 012740 mov $5,-(r0) 000005 105710 1: tstb (r0) 002376 bge 1b 005007 clr pc RP (drive 0) 012700 mov $rpmr,r0 176726 005040 clr -(r0) 005040 clr -(r0) 005040 clr -(r0) 010040 mov r0,-(r0) 012740 mov $5,-(r0) 000005 105710 1: tstb (r0) 002376 bge 1b 005007 clr pc TM (drive 0): 012700 mov $tmba,r0 172526 010040 mov r0,-(r0) 012740 mov $60003,-(r0) 060003 000777 br . FILES
/unix system code /boot system bootstrap /etc/init system process dispatcher /mdec/xxuboot sector 0 boot blocks, xx is disk type SEE ALSO
crash(8V), autoconfig(8), reboot(2), disklabel(8), fsck(8), init(8) 3rd Berkeley Distribution May 24, 1996 BOOT(8)
All times are GMT -4. The time now is 07:23 PM.
Unix & Linux Forums Content Copyright 1993-2022. All Rights Reserved.
Privacy Policy