kill / pkill / killall

Use kill to send a signal by specifying a PID.

kill PID
kill -signal_name PID
kill -signal_number PID

Use pkill to search by process name and send a signal.

pkill process_name
pkill -signal process_name
pkill -u username process_name

Use killall to send a signal to all processes with an exactly matching name.

killall process_name
killall -signal process_name

List available signals.

kill -l

The following examples assume that sleep 1000 is running with PID 12345.

ps aux | grep "sleep" | grep -v grep
alice  12345  0.0  0.0  5536  1024 pts/0  S  10:00  0:00 sleep 1000

Request graceful termination with SIGTERM. When no signal is specified, the default SIGTERM (15) is sent.

kill 12345

Use SIGKILL (-9) to force-terminate an unresponsive process. The process stops immediately without any cleanup.

kill -9 12345
kill -SIGKILL 12345   # same effect

Send SIGHUP to nginx to reload its configuration file. Most daemons reload their config when they receive this signal.

kill -HUP $(cat /var/run/nginx.pid)

Use pkill to stop a process by name, saving you the trouble of looking up its PID.

pkill sleep          # send SIGTERM to processes whose name contains "sleep"
pkill -9 myapp       # force-terminate
pkill -u www-data php-fpm   # stop a process owned by a specific user

killall sends a signal to all processes whose name matches exactly.

killall nginx

Use kill -0 to check whether a process exists without sending an actual signal. This is useful for monitoring processes in scripts.

sample_check_process.sh

kill 12345 && echo "Signal sent successfully"
sleep 1
if ! kill -0 12345 2>/dev/null; then
    echo "Process has terminated"
fi

Run the following command:

bash check_process.sh
Signal sent successfully
Process has terminated

Example of capturing a background process PID and stopping it.

sample_kill_bg.sh

sleep 100 &
bg_pid=$!
echo "Background PID: $bg_pid"
kill "$bg_pid"
echo "Stop signal sent"

Run the following command:

bash kill_bg.sh
Background PID: 12348
Stop signal sent

Common Mistake 1: Sending SIGKILL before trying SIGTERM

SIGKILL cannot be caught or ignored, so the process has no chance to clean up. Sending it first can cause data corruption or leave lock files behind.

kill -9 12345
（process has no chance to flush data or release resources）

Send SIGTERM first and wait for the process to exit gracefully. Use SIGKILL only if it does not respond.

kill 12345
sleep 3
kill -0 12345 2>/dev/null && kill -9 12345

Common Mistake 2: pkill pattern matches unintended processes

pkill matches any process whose name contains the pattern. A short pattern can match processes you did not intend to kill.

pkill php
（may kill php-fpm, php-cgi, and any other process with "php" in the name）

Use -x to match the full process name exactly.

pkill -x php

The recommended practice is to first send SIGTERM (-15) to give the process a chance to exit gracefully, and only use SIGKILL (-9) if the process does not terminate. SIGKILL skips flushing open files and releasing resources, which can result in data corruption.

$ kill -0 PID only checks whether the process exists without sending any signal (exit status 0 = exists, non-zero = does not exist). This is useful for process monitoring in scripts.

To inspect processes, see ps / top / htop. For running processes in the background, see & (background execution).