RH134 Ch 3: tuned Profiles and Process Priority

tuned adjusts kernel and device settings for a workload by applying a profile. The tuned service must be enabled so the profile is reapplied after every boot.

bash systemctl enable --now tuned tuned-adm list available profiles tuned-adm active current profile tuned-adm recommend what tuned suggests for this machine tuned-adm profile throughput-performance tuned-adm off stop tuning

Useful profiles: balanced (general purpose), throughput-performance (servers), latency-performance, virtual-guest (VMs; builds on throughput-performance), virtual-host and powersave. Custom profiles live in /etc/tuned/<name/tuned.conf and can include= an existing one.

Process priority. The Linux scheduler shares CPU time according to each process's nice value, from -20 (greediest) to 19 (most polite); the default is 0. Anyone may make their own processes nicer; only root can lower the value.

bash nice -n 15 ./archiver & start with nice 15 renice -n 10 -p 2345 change a running process ps -o pid,ni,pcpu,comm -C archiver see the NI column top press r to renice interactively

Find the process first with pgrep -x name or ps aux --sort=-pcpu | head.

Common exam mistakes - Changing the profile while tuned is disabled, so nothing reapplies it after reboot. - Using nice on a running process: nice only starts new ones; use renice (with -n) for existing PIDs. - Confusing priority (PR in top) with the nice value (NI). Tasks normally ask for NI. - Starting a background job that dies when you log out: use nohup or a systemd unit.