htop explained

The page explains what htop and top display on Linux by tracing each field back to the kernel and /proc. It covers uptime, load average, process counts, PIDs, process trees, owners, states, time sharing, niceness, memory columns, and startup services.

Reading notes

#

• Uptime comes from /proc/uptime, and uptime formats that data for humans.
• Load average comes from /proc/loadavg, where the first three numbers reflect 1, 5, and 15 minute periods.
• Load average counts running and uninterruptible processes, so it is not the same as CPU usage.
• htop shows tasks, threads, and kernel threads, and those views can be toggled with keyboard shortcuts.
• PIDs identify processes, and /proc/<pid>/ exposes command line, current directory, executable link, and other process details.
• Process hierarchies come from parent and child relationships created by fork and exec.
• Process ownership is based on user IDs, with names resolved through NSS and files such as /etc/passwd, /etc/group, and /etc/shadow.
• sudo and setuid binaries change which user a command runs as.
• Process states include running, interruptible sleep, uninterruptible sleep, zombie, stopped, and traced.
• Signals such as SIGINT, SIGTERM, SIGKILL, STOP, and CONT are used to interrupt, stop, resume, or kill processes.
• Time slicing lets one CPU appear to run several processes at once.
• Niceness affects scheduler priority, and the article relates that to htop color indicators.
• Memory columns in htop distinguish virtual memory, resident memory, shared memory, and memory percentage.
• The post also lists several Ubuntu server services and discusses which ones might be unnecessary on a minimal machine.
• The appendix shows how to use strace, dpkg -S, and source code to learn what programs do.