Commands using kill (107)

  • this exits bash without saving the history. unlike explicitly disabling the history in some way, this works anywhere, and it works if you decide *after* issuing the command you don't want logged, that you don't want it logged ... $$ ( or ${$} ) is the pid of the current bash instance this also works perfectly in shells that don't have $$ if you do something like kill -9 `readlink /proc/self`

    kill -9 $$
    stu · 2009-03-27 23:13:53 17

  • 12
    find /proc -user myuser -maxdepth 1 -type d -mtime +7 -exec basename {} \; | xargs kill -9
    sharfah · 2009-10-05 14:49:51 3

  • 12
    kill -9 -1
    cogsmos · 2010-11-08 03:28:43 2
  • It identifies the parents of the Zombie processes and kill them. So the new parent of orphan Zombies will be the Init process and he is already waiting for reaping them. Be careful! It may also kill your useful processes just because they are not taking care and waiting for their children (bad parents!). Show Sample Output

    kill -9 `ps -xaw -o state -o ppid | grep Z | grep -v PID | awk '{print $2}'`
    khashmeshab · 2010-10-27 07:29:14 9
  • I wanted to create a copy of my whole laptop disk on an lvm disk of the same size. First I created the logical volume: lvcreate -L120G -nlaptop mylvms SOURCE: dd if=/dev/sda bs=16065b | netcat ip-target 1234 TARGET: nc -l -p 1234 | dd of=/dev/mapper/mylvms-laptop bs=16065b to follow its process you issue the following command in a different terminal STATS: on target in a different terminal: watch -n60 -- kill -USR1 $(pgrep dd) (see

    SOURCE: dd if=/dev/sda bs=16065b | netcat ip-target 1234 TARGET: netcat -l -p 1234 | dd of=/dev/mapper/laptop bs=16065b STATS on target: watch -n60 -- kill -USR1 $(pgrep dd)
    bw · 2009-12-16 10:51:06 4
  • This command seems to achieve the similar/same goal.

    kill -l
    b_t · 2010-11-17 23:48:51 0
  • or "Execute a command with a timeout" Run a command in background, sleep 10 seconds, kill it. ! is the process id of the most recently executed background command. You can test it with: find /& sleep10; kill $!

    very_long_command& sleep 10; kill $!
    dooblem · 2010-04-29 20:43:13 1

  • 5
    kill -9 `ps -u <username> -o "pid="`
    skye · 2009-03-23 17:51:37 6
  • Send signal 0 to the process. The return status ($?) can be used to determine if the process is running. 0 if it is, non-zero otherwise.

    kill -0 [pid]
    sharfah · 2009-05-19 11:37:20 4
  • This is a more accurate way to watch the progress of a dd process. The $DDPID=$! is needed so that you don't get the PID of the sleep. The sleep 1 is needed because in my testing at least, if you run kill -USR1 against dd too quickly, it will kill it off instead of display the status. So you need to wait a second, probably so that it can configure itself to trap the USR1 signal. Show Sample Output

    dd if=fromfile of=tofile & DDPID=$! ; sleep 1 ; while kill -USR1 $DDPID ; do sleep 5; done
    deltaray · 2010-01-12 15:01:44 1
  • I like much more the perl solution, but without using perl. It launches a backgroup process that will kill the command if it lasts too much. A bigger function: check_with_timeout() { [ "$DEBUG" ] && set -x COMMAND=$1 TIMEOUT=$2 RET=0 # Launch command in backgroup [ ! "$DEBUG" ] && exec 6>&2 # Link file descriptor #6 with stderr. [ ! "$DEBUG" ] && exec 2> /dev/null # Send stderr to null (avoid the Terminated messages) $COMMAND 2>&1 >/dev/null & COMMAND_PID=$! [ "$DEBUG" ] && echo "Background command pid $COMMAND_PID, parent pid $$" # Timer that will kill the command if timesout sleep $TIMEOUT && ps -p $COMMAND_PID -o pid,ppid |grep $$ | awk '{print $1}' | xargs kill & KILLER_PID=$! [ "$DEBUG" ] && echo "Killer command pid $KILLER_PID, parent pid $$" wait $COMMAND_PID RET=$? # Kill the killer timer [ "$DEBUG" ] && ps -e -o pid,ppid |grep $KILLER_PID | awk '{print $1}' | xargs echo "Killing processes: " ps -e -o pid,ppid |grep -v PID | grep $KILLER_PID | awk '{print $1}' | xargs kill wait sleep 1 [ ! "$DEBUG" ] && exec 2>&6 6>&- # Restore stderr and close file descriptor #6. return $RET }

    $COMMAND 2>&1 >/dev/null & WPID=$!; sleep $TIMEOUT && kill $! & KPID=$!; wait $WPID
    keymon · 2010-05-26 11:12:26 0
  • Somtime one wants to kill process not by name of executable, but by a parameter name. In such cases killall is not suitable method.

    kill -9 `ps ax | egrep [f]elix.jar | egrep -o -e '^ *[0-9]+'`
    yababay · 2010-09-30 16:45:47 0
  • This is a 'killall' command equivalent where it is not available. Prior to executing it, set the environment variable USERNAME to the username, whose processes you want to kill or replace the username with the $USERNAME on the command above. Side effect: If any processes from other users, are running with a parameter of $USERNAME, they will be killed as well (assuming you are running this as root user) [-9] in square brackets at the end of the command is optional and strongly suggested to be your last resort. I do not like to use it as the killed process leaves a lot of mess behind.

    ps -ef | grep $USERNAME | awk {'print $2'} | xargs kill [-9]
    TheNomad · 2009-02-17 20:35:35 3
  • Logs all users out except root. I changed the grep to use a regexp in case a user's username contained the word root.

    who -u | grep -vE "^root " | kill `awk '{print $7}'`
    ProfessorTux · 2010-11-05 17:43:41 19
  • Adjust "sleep X" to your needs. *NOTE: First sleep is required because bash doesn't have a "post-test" syntax (do XXX while). Show Sample Output

    dd if=/path/to/inputfile of=/path/to/outputfile & pid=$! && sleep X && while kill -USR1 $pid; do sleep X; done
    cyrusza · 2010-12-02 15:07:18 3
  • alternatively, run the spinner for 5 seconds: timeout 5 bash -c 'spinner=( Ooooo oOooo ooOoo oooOo ooooO oooOo ooOoo oOooo); while true; do for i in ${spinner[@]}; do for j in seq 0 ${#i}; do echo -en "\b\b"; done; echo -ne "${i}"; sleep 0.2; done; done' Show Sample Output

    while kill -0 0; do timeout 5 bash -c 'spinner=( Ooooo oOooo ooOoo oooOo ooooO oooOo ooOoo oOooo); while true; do for i in ${spinner[@]}; do for _ in seq 0 ${#i}; do echo -en "\b\b"; done; echo -ne "${i}"; sleep 0.2; done; done'; done
    anapsix · 2015-05-07 19:13:08 4

  • 2
    for i in `ps aux | grep ssh | grep -v grep | awk {'print $2'}` ; do kill $i; done
    jcwaters · 2009-02-21 02:07:22 7

  • 2
    kill HUP `pidof '/usr/bin/X'`
    aceiro · 2010-07-16 03:55:13 1
  • This shell function takes a single argument, which is used as the base name of the .wav, .timing and .session files created. To create a screencast: screencast test type and talk ... then type 'exit' or to exit the screencast. test.wav will contain the audio from your screencast. test.session will contain text and control characters needed to paint the screen test.timing will contain timing information needed to synch individual keystrokes in test.session with the audio. to play back: aplay test.wav & scriptreplay test.{timing,session} NOTE: because the shell function uses the variable "$!", and bash likes to expand '!' during history expansion, you will need to turn off bash's history before you enter the shell function. This can be achieved using the command set +H

    screencast() { arecord -R 1000 -f cd -t wav $1.wav & RECPID=$!; echo "Starting screencast in new shell. Exit subshell to quit."; script -t 2> $1.timing -a $1.session; kill $RECPID; }
    bartonski · 2011-01-20 14:35:47 2
  • Only slightly different than previous commands. The benefit is that your "watch" should die when the dd command has completed. (Of course this would depend on /proc being available)

    dd if=/dev/urandom of=file.img bs=4KB& pid=$!; while [[ -d /proc/$pid ]]; do kill -USR1 $pid && sleep 1 && clear; done
    unixmonkey14934 · 2011-06-24 21:49:10 4
  • Kill all processes with foo in them. Similar to pkill but more complete and also works when there is no pkill command. Works on almost every Linux/Unix platform I have tried.

    ps -ef | grep [f]oo | awk '{print $2}' | xargs kill -9
    eredicatorx · 2009-02-05 13:43:01 22
  • explanation: grep -- displays process ids -v -- negates the matching, displays all but what is specified in the other options -u -- specifies the user to display, or in this case negate The process loops through all PIDs that are found by pgrep, then orders a forced kill to the processes in numerical order, effectively killing the parent processes first including the shells in use which will force the users to logout. Tested on Slackware Linux 12.2 and Slackware-current

    for i in $(pgrep -v -u root);do kill -9 $i;done
    lostnhell · 2009-03-24 02:54:52 4
  • USAGE: gate listening_port host port Creates listening socket and connects to remote device at host:port. It uses pipes for connection between two sockets. Traffic which goes through pipes is wrote to stdout. I use it for debug network scripts.

    gate() { mkfifo /tmp/sock1 /tmp/sock2 &> /dev/null && nc -p $1 -l < /tmp/sock1 | tee /tmp/sock2 & PID=$! && nc $2 $3 < /tmp/sock2 | tee /tmp/sock1; kill -KILL $PID; rm -f /tmp/sock1 /tmp/sock2 ; }
    true · 2009-09-25 08:10:23 1
  • Add that and "cont () { ps -ec | grep $@ | kill -SIGCONT `awk '{print $1}'`; }" (without the quotes) to you bash profile and then use it to pause and resume processes safely

    stop () { ps -ec | grep $@ | kill -SIGSTOP `awk '{print $1}'`; }
    iridium172 · 2009-12-27 19:40:09 4
  • Another way of counting the line output of tail over 10s not requiring pv. Cut to have the average per second rate : tail -n0 -f access.log>/tmp/tmp.log & sleep 10; kill $! ; wc -l /tmp/tmp.log | cut -c-2 You can also enclose it in a loop and send stderr to /dev/null : while true; do tail -n0 -f access.log>/tmp/tmp.log & sleep 2; kill $! ; wc -l /tmp/tmp.log | cut -c-2; done 2>/dev/null

    tail -n0 -f access.log>/tmp/tmp.log & sleep 10; kill $! ; wc -l /tmp/tmp.log
    dooblem · 2010-04-29 21:23:46 12
  •  1 2 3 >  Last ›

What's this? is the place to record those command-line gems that you return to again and again. That way others can gain from your CLI wisdom and you from theirs too. All commands can be commented on, discussed and voted up or down.

Share Your Commands

Check These Out

exim statistics about mails from queue
statistics are sorted based on number of recipients.

Setting reserved blocks percentage to 1%
According to tune2fs manual, reserved blocks are designed to keep your system from failing when you run out of space. Its reserves space for privileged processes such as daemons (like syslogd, for ex.) and other root level processes; also the reserved space can prevent the filesystem from fragmenting as it fills up. By default this is 5% regardless of the size of the partition.

Simple Video Surveillance by email
This takes a picture (with the web cam) every 5 minutes, and send the picture to your e-mail. Some systems support mail -a "References: " so that all video surveillance emails are grouped in a single email thread. To keep your inbox clean, it is still possible to filter and move to trash video surveillance emails (and restore these emails only if you really get robbed!) For instance with Gmail, emails sent to can be filtered with "Matches:"

Install pip with Proxy
Installs pip packages defining a proxy

create tar archive of files in a directory and its sub-directories
creates a compressed tar archive of files in /path/foo and writes to a timestamped filename in /path.

Convert CSV to JSON
Replace 'csv_file.csv' with your filename.

Use AWS CLI and JQ to get a list of instances sorted by launch time
You can do the filtering natively in the aws cli, without using jq (although jq is awesome!)

Rsync remote data as root using sudo
If your user has sudo on the remote box, you can rsync data as root without needing to login as root. This is very helpful if the remote box does not allow root to login over SSH (which is a common security restriction).

reset an hanging terminal session
when your terminal session seems unrensponsive (this normally happen after outputting some binary data directly on your standard output) it may me saned by hitting: CTRL+J tput sgr0 CTRL+J Note: don't press the Enter key, just ctrl+j

Convert seconds to [DD:][HH:]MM:SS
Converts any number of seconds into days, hours, minutes and seconds. sec2dhms() { declare -i SS="$1" D=$(( SS / 86400 )) H=$(( SS % 86400 / 3600 )) M=$(( SS % 3600 / 60 )) S=$(( SS % 60 )) [ "$D" -gt 0 ] && echo -n "${D}:" [ "$H" -gt 0 ] && printf "%02g:" "$H" printf "%02g:%02g\n" "$M" "$S" }

Stay in the loop…

Follow the Tweets.

Every new command is wrapped in a tweet and posted to Twitter. Following the stream is a great way of staying abreast of the latest commands. For the more discerning, there are Twitter accounts for commands that get a minimum of 3 and 10 votes - that way only the great commands get tweeted.


Subscribe to the feeds.

Use your favourite RSS aggregator to stay in touch with the latest commands. There are feeds mirroring the 3 Twitter streams as well as for virtually every other subset (users, tags, functions,…):

Subscribe to the feed for: