Processes

Every program you run on Unix becomes a process. The kernel keeps track of all of them — who owns them, how much memory they are using, how long they have been running. When something goes wrong, the first thing you do is look at the process list.

Viewing processes with ps

ps stands for "process status." Run without options it only shows processes attached to your current terminal. To see everything, use ps aux — a, u, and x mean "all users," "detailed format," and "include processes without a terminal."

$ ps aux
USER       PID %CPU %MEM    VSZ   RSS TT  STAT STARTED    TIME COMMAND
alice     1234  0.5  2.3 423456 23456 ??  S    10:15AM  0:02.32 nginx
root      1235  0.0  0.1 123456  1234 ??  Ss   10:15AM  0:00.01 nginx
alice     5678  1.2  8.1 1023456 89123 ??  R    10:20AM  0:15.42 node server.js

The PID column is the process ID — unique number for each running process. The COMMAND column shows what is running. %CPU and %MEM tell you how much of each resource it is consuming.

Interactive monitoring with top

top shows a live, updating view of processes sorted by CPU usage. It refreshes every few seconds. Hit q to quit. If you have htop installed, it is even nicer — color-coded and scrollable.

$ top
top - 10:25:01 up 3 days,  2:15,  2 users,  load average: 0.45, 0.32, 0.20
Tasks: 123 total,   1 running, 122 sleeping
%Cpu(s):  8.2 us,  2.1 sy, 89.7 id
KiB Mem :  8123456 total,  2345678 free

Killing processes with kill

When a process freezes or consumes too much, you can terminate it with kill. You need to know the PID. kill PID sends SIGTERM (signal 15) — a polite request to shut down. Most processes comply.

$ kill 5678

If a process refuses to die, use kill -9 PID to send SIGKILL. It cannot be ignored. Think of it as the difference between asking someone to leave and throwing them out.

$ kill -9 5678

Be careful with kill -9. The process gets no chance to clean up, which can corrupt files or leave things in a bad state. Use the regular kill first.