Don't Panic - Linux Process Management

A Linux process is a running program represented by kernel-managed state: an identifier, credentials, memory mappings, open files, signal settings, scheduling attributes, and one or more threads. The scheduler selects runnable threads for CPUs. User-space tools inspect and change that state through system calls.

Process management is a control loop: identify the correct task, inspect its state and relationships, choose the least disruptive action, then verify the result. A PID is useful until reuse after exit. Names are ambiguous when several processes share an executable. Act on relationships such as PPID, process groups, and cgroups when identity is unclear.

The common lifecycle uses fork() (or related calls) to create a child with a separate address space, then execve() to replace the image while keeping the PID. Parents and children form a tree. Process groups and sessions organize signal delivery and terminal job control. Signals request actions; some can be caught or ignored, and some terminate immediately.

Scheduling attributes and resource controls decide which threads run when CPUs contend. Zombies remain until a parent reaps them and can exhaust PID space without burning CPU. Uninterruptible sleep usually points at I/O or kernel waits rather than a signal you can send from the shell.

Read the Intro for the lifecycle and identity rules. Use the Cheatsheet when you need the signal and scheduler maps. Landscape and Timeline place process tools among related Linux administration topics; Updates tracks kernel.org releases that change these interfaces. Verify after every signal or priority change.

Don't Panic - Linux Process Management

Relevant careers