Priorytety procesów za pomocą nice, renice i sygnałów

You already know how to view processes with ps/top and stop them with kill. But sometimes a process should simply run more politely so it does not starve others of CPU.

That is what nice and renice are for.

Every process has a niceness from -20 (highest priority) to 19 (lowest priority). A nicer process yields CPU to others. Default is 0.

nice -n N command launches a command with niceness N. Higher numbers are friendlier to the rest of the system.

The NI column in top shows niceness. With ps you can request it explicitly.

renice adjusts the niceness of an already running process by PID.

Any user can make a process nicer (raise the number). Only root can give a process a negative niceness, because that takes CPU away from everyone else.

The kill command really sends a signal. SIGTERM (15) asks nicely to stop; SIGKILL (9) forces it; SIGHUP (1) often tells a daemon to reload config.

Sending SIGHUP to many services makes them re-read their configuration without a full restart, avoiding downtime.

SIGSTOP freezes a process and SIGCONT resumes it, handy for temporarily relieving load.

pkill and killall let you send signals by process name instead of PID.

A typical admin flow: spot a CPU hog in top, lower its priority with renice instead of killing it, and only escalate to kill -9 if it misbehaves.

Which niceness value gives a process the highest CPU priority?

You learned to fine-tune process behavior:

• nice -n N cmd starts at a chosen priority
• renice changes priority of running PIDs
• Negative niceness needs root
• Signals like SIGTERM, SIGHUP, SIGSTOP/SIGCONT control processes without killing them