How to kill a process

The easiest way: use pkill to kill a running process by name.

Let’s say you have a background task running:

Those sleep commands will run for 60 seconds and 600 seconds, respectively.

Kill a process by name

If we want to kill them sooner, we can just pkill them by name:

pkill works similar to killall—both commands will kill all matching processes. What if we only want to kill one?

Kill a process by ID

Maybe you have multiple processes running with similar names, and you only want to kill one of them. In that case, you can run kill {pid}.

In the example above, pgrep returned the process IDs for both sleep commands:

To stop just one of them I could have run kill 26912, and left the other one untouched. But which PID is which? They’re going to run for different amounts of time, and I want to make sure I kill the right one.

How to find process IDs

In the screenshot above, pgrep didn’t differentiate between the two sleep commands; it just gave us the PIDs. The traditional way to find PIDs is via ps aux; specifically: ps aux | grep {name_to_look_for}.

However, I find pgrep -lf  to be a nicer option¹:

Now we see the processes with their details, and can decide what to do from here.

What if pkill and kill don’t work?

By default, pkill and kill send processes a SIGTERM, which is the signal to terminate gracefully. Sometimes though, that doesn’t work. When it doesn’t, you may need to send a SIGKILL via kill -9 {PID} or pkill -9 {name}.

Sending SIGKILL does carry some risks, so you shouldn’t do it unless you know it’s ok. For more details, see this helpful article.

---

Helpful Links

• The differences between pkill and killall – Stack Exchange
• SIGTERM vs SIGKILL

Notes

1. -l is “long output”, which gives the process name in addition to its PID e.g. 48289 sleep. When combining -l and -f we get the process’s full command e.g. 48289 sleep 100 ↩︎