Under unix-like systems, the
kill command along with
killall are used regularly to stop a process and tell it to close down. Under these systems you can use
killall to send “signals” to any running process that you have permission to access (usually those which you run under your own login ID). A signal is like an interrupt to the cpu only for processes. There are many types of signal defined by POSIX, but the two common ones used with
SIGHUP (hangup), and
SIGKILL (terminate immediately).
These two signals tell the program to close down neatly and cease operation in the case of
SIGHUP, and with
SIGKILL tell the kernel to immediately destroy the process without waiting for the program to tidy up after itself.
SIGHUP is the default signal sent by
killall, while the
SIGKILL can be sent via the
-KILL commandline switch to said commands.
SIGKILL is useful for when an application is completely unresponsive to input, and needs to be forcibly closed as it won’t respond to any
SIGHUP signals to close cleanly due to unresponsiveness.
The difference between
killall is the processes that they enact the signal upon.
kill fires the signal at a single process identified by the numeric Process ID which is specified on the commandline.
killall fires the signal at any and all processes which match the executable name, such as
firefox-bin in the case of the popular web browser. As with
kill the executable name is a commandline argument to the
killall program. Example usage follows:
Kills the process with PID of
12345 by sending the
SIGHUP to get the program to shut itself down cleanly.
Sends all firefox processes the
SIGHUP signal to get them all to shutdown cleanly.
Explicitly terminates the process with ID
54321 without waiting for it to clean up after itself.
Explicitly terminates all thunderbird processes without waiting for them to clean up after themselves.
In the windows world, the concept of signals isn’t so apparent, and so termination of processes can’t be done by sending a
SIGHUP signal to the appropriate image. (Image is the windows term for an executable.) The traditional way to stop a program in Windows is to use the GUI taskmanager. However this is not useful for times when you are logged into a Windows machine via telnet or SSH.
Telnet and SSH are commandline protocols only, and therefore any GUI applications are going to be completely unavailable to the administrator logged into a Windows machine via these methods. Instead, Microsoft has provided the
taskkill commandline programs which take the place of POSIX’s
Tasklist dumps a list of all running tasks on the system into the console window (or over the Telnet/SSH link). From this list you can determine the image names and process IDs of any task currently active. Next is the
taskkill program, which as the name suggests is a way of terminating the execution of a task (or many tasks as we’ll explore). Taskkill can take two different methods of identifying the process(es) to be killed. Examples below:
Kill all processes with the image name of
Kill the single process with the PID of
Finally, one can append the
/F switch to the commandline which will change the symantics of how the system will kill the identified process(es). Without the
taskkill acts analogously to
killall or more explicitly
kill -HUP or
killall -HUP. When the
/F switch is added to the commandline, taskkill acts more like
kill -KILL or
I discovered this tip on the tweaks.com article about
taskkill. Their article goes into more depth about the Windows side of things, while I have tried to explain the similarities between the Windows and POSIX paradigms. Be sure to check out tweaks.com for more tips and tricks for Windows, and specifically the article linked above for more information about the
taskkill filter mechanism which allows much more fine-grained control over which process you wish to kill.