Skip to content

README.md

Latest commit

 

History

History
213 lines (159 loc) · 9.01 KB

README.md

File metadata and controls

213 lines (159 loc) · 9.01 KB
Raw

htop — The Complete Reference

An interactive, colorized, real-time process viewer A friendlier, more visual alternative to top — same underlying idea (live, continuously refreshing process monitoring), much easier to use.

Table of Contents

What is htop?

htop is an interactive, full-screen process viewer that continuously refreshes to show what's running on the system right now — CPU and memory usage per core, per process, with color-coded bars, mouse support, and the ability to sort, filter, search, and send signals directly from the interface, without ever leaving it.

htop
# Launches the full-screen interactive interface — no useful output
# when piped or redirected, since it's designed for live interaction

Why people reach for htop over top: top's interface is functional but famously unfriendly (obscure single-letter commands, plain text, no mouse support by default). htop presents the same fundamental information — process list, resource usage, load — with color-coded per-core CPU bars, a visible tree view of parent/child relationships, and mouse-clickable columns/processes, at the cost of being a separate package that isn't always preinstalled.

Where does htop live?

/usr/bin/htop

which htop
htop --version
# htop 3.3.0

Unlike ps, top, and kill (all part of core packages present on virtually every Linux system), htop is a separate, third-party package and is not preinstalled on most distributions by default:

sudo apt install htop      # Debian/Ubuntu
sudo dnf install htop      # Fedora/RHEL
sudo pacman -S htop        # Arch
brew install htop          # macOS (via Homebrew)

How htop works internally

Like ps and top, htop reads process and system information from the /proc filesystem on Linux — it doesn't have privileged kernel access beyond what any user-space program reading /proc can see. It re-reads this data on a fixed interval (configurable, roughly once per second by default) and redraws the entire screen with the refreshed values.

ls /proc/1234/
# stat  status  cmdline  cwd  fd  maps  ...
# htop draws its process rows, CPU%, and memory columns from exactly
# these same kernel-exposed files that ps and top also read.

Because htop is a separate process itself, watching htop naturally includes htop's own (typically very small) resource footprint in the very list it's displaying — a minor but sometimes noticed self-referential quirk.

Syntax

htop [OPTIONS]

Nearly all of htop's real functionality is interactive, driven by keyboard/mouse input once it's running, rather than through command-line flags — the flags mostly control startup behavior (which user's processes to show, sort order, delay).

Understanding the Screen

  1  [||||||||||||        45.2%]   Tasks: 87, 312 thr; 3 running
  2  [||||               12.1%]   Load average: 1.25 0.98 0.87
  Mem[|||||||||||       4.2G/16G]  Uptime: 2 days, 04:12:33
  Swp[                   0K/2G]

  PID USER      PRI  NI  VIRT   RES   SHR S CPU% MEM%   TIME+  Command
 1234 alice      20   0  812M  145M  42M  R 23.4  0.9  1:32.10 firefox
 5678 alice      20   0   22M   9M   6M  S  0.3  0.1  0:00.42 bash
Section Meaning
Per-core CPU bars One bar per logical CPU core, color-coded by usage type (see below)
Mem bar Physical RAM used vs. total
Swp bar Swap space used vs. total
Tasks Total process/thread count, and how many are actively running right now
Load average Same 1/5/15-minute figures reported by uptime
Process table Sortable, scrollable list of individual processes — same core fields as ps, refreshed continuously

The Header Meters — CPU, Memory, Swap, Load

The per-core CPU bars use color segments to break down what kind of activity is consuming CPU, not just how much:

Color (default theme) Meaning
Green Normal user-space processes
Red Kernel (system) time
Blue Low-priority ("niced") processes
Orange/Yellow I/O wait (varies by htop version/theme)
Cyan/Magenta Virtualization steal time or IRQ time, depending on version

This breakdown is one of htop's genuine advantages over plain top's aggregate percentage — at a glance, a fully "red" bar signals heavy kernel-side activity, quite different from a fully "green" one, without needing to separately run something like vmstat.

Keyboard Controls

Key Action
F1 / h Help screen
F2 / S Setup — customize meters, columns, colors, display options
F3 / / Search/incremental filter for a process by name
F4 Filter the list to only matching processes
F5 / t Toggle tree view (parent/child process hierarchy)
F6 Change the sort column
F9 / k Kill the selected process (choose a signal from a menu)
F7 / ] Increase selected process's priority (renice, lower number)
F8 / [ Decrease selected process's priority (renice, higher number)
Space Tag/select a process (for acting on multiple at once)
U Untag all tagged processes
F5 Toggle tree view
F10 / q Quit

Mouse support is enabled by default in most terminal emulators — clicking a column header sorts by it, and clicking a process selects it.

Sorting, Filtering, and Searching

# Interactively, inside htop:
# F6 (or click a column header) — choose which column to sort by
# > or < — cycle sort column without opening the menu
# F4 — type a substring to FILTER the visible list down to matches
# F3 or / — SEARCH and jump to the next matching process without hiding others

# From the command line, at startup:
htop -s PERCENT_CPU     # start already sorted by CPU usage
htop -u alice           # show only processes owned by user alice
htop -p 1234,5678       # show only these specific PIDs

All Key Command-Line Options

Option Long form Description
-d SECONDS --delay=SECONDS Set the refresh interval (in tenths of a second)
-u USER --user=USER Show only processes owned by the given user
-p PID,... --pid=PID,... Show only the specified PID(s)
-s COLUMN --sort-key=COLUMN Start already sorted by the given column
-t --tree Start in tree view
-C --no-color Disable color output
-n NUMBER --iterations=NUMBER Exit automatically after N updates (useful for scripted, bounded runs)
— --version Print version
— --help Print usage help

htop vs top vs ps vs glances

Tool Best for Key difference from htop
top Live view, guaranteed present on virtually every Linux system Plain text, less discoverable controls, no mouse support by default, no built-in tree view toggle in older versions
ps One-off, scriptable snapshots Not live/interactive at all — prints once and exits, ideal for piping into other commands
htop Live, human-friendly interactive monitoring The most approachable of the interactive viewers, at the cost of not being preinstalled everywhere
glances Broader system overview (network, disk I/O, sensors, containers) alongside processes Covers more subsystems in one dashboard than htop's process-centric focus 
ps aux --sort=-%cpu | head    # scripted, one-shot, pipeable
top                            # live view, always available, plain
htop                           # live view, friendlier, needs installing
glances                        # live view, broader system-wide dashboard

Related Commands

Command Relation
top The original live process viewer htop is modeled after and improves on
ps Non-interactive, scriptable snapshot equivalent
glances Broader live system-monitoring dashboard covering more than just processes
iotop htop-like live view specifically for per-process disk I/O
nmon / dstat Other live system-resource monitoring alternatives
kill / renice What htop's F9/F7/F8 keys invoke under the hood

See also: examples.md · edge-cases.md · interview-questions.md