Exercise: Find the slowdown and tune the server

Find the slowdown and tune the server

• Read load average, top and free.
• Compare a normal and a nice process.
• Install tuned and switch profiles.

Start the exercise

On workstation, start the exercise. It makes sure the tuned package is not installed yet on servera.

[student@workstation ~]$ lab start sa-monitoring

Later tasks assume that you are logged in to the server they name: ssh student@servera from workstation, then sudo -i for a root shell.

Take the baseline

On servera, record the load average, the number of CPUs and the memory figures before you start any load.

[student@workstation ~]$ ssh student@servera
[student@servera ~]$ uptime
 16:21:40 up 0 min,  0 users,  load average: 0.00, 0.01, 0.00
[student@servera ~]$ nproc
1
[student@servera ~]$ free -h | head -2
               total        used        free      shared  buff/cache   available
Mem:           937Mi       376Mi       553Mi        21Mi       166Mi       560Mi

Start two hogs, one of them nice

Start one sha256sum /dev/zero normally and one with nice -n 19, both in the background. After a few seconds, compare their CPU shares and nice values with ps.

[student@servera ~]$ sha256sum /dev/zero &
[1] 5062
[student@servera ~]$ nice -n 19 sha256sum /dev/zero &
[2] 5063
[student@servera ~]$ ps -o pid,ni,pcpu,stat,cmd -C sha256sum
    PID  NI %CPU STAT CMD
   5062   0 95.0 R    sha256sum /dev/zero
   5063  19  1.3 RN   sha256sum /dev/zero

Both want all the CPU. The normal one gets almost all of it; the nice-19 one only runs when there is a gap. The N in STAT marks a low-priority process.

Try to give the nice process more priority

Try renice -n 10 on the nice-19 process, as student. What happens? Then make the normal process nicer instead (renice -n 5).

[student@servera ~]$ renice -n 10 -p 5063
renice: failed to set priority for 5063 (process ID): Permission denied
[student@servera ~]$ renice -n 5 -p 5062
5062 (process ID) old priority 0, new priority 5

Making a process less nice needs root, so users cannot raise their own priority. Making a process nicer is always allowed.

Watch the load rise

With both hogs running, find them with top (press q to leave) and look at the header. Wait a minute and run uptime again. Which of the three averages moved first?

[student@servera ~]$ top -b -n1 -o %CPU | sed -n 1,3p
top - 16:29:34 up 8 min,  0 users,  load average: 0.84, 0.47, 0.22
Tasks: 121 total,   2 running, 119 sleeping,   0 stopped,   0 zombie
%Cpu(s): 98.0 us,  2.0 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
[student@servera ~]$ uptime
 16:30:34 up 9 min,  0 users,  load average: 1.82, 0.97, 0.43

The 1-minute figure reacts first, then 5 and 15. With two hogs on one CPU the load climbs towards 2. Your exact numbers will differ.

End the load

[student@servera ~]$ pkill sha256sum
[student@servera ~]$ jobs
[1]-  Terminated              sha256sum /dev/zero
[2]+  Terminated              nice -n 19 sha256sum /dev/zero

Install tuned and choose a profile

As root, install the tuned package, enable and start the service, and show the active and recommended profiles. Switch to throughput-performance, confirm, and then switch back to the recommended one.

[student@servera ~]$ sudo dnf install -y tuned
...output omitted...
Complete!
[student@servera ~]$ sudo systemctl enable --now tuned
[student@servera ~]$ tuned-adm active
Current active profile: virtual-guest
[student@servera ~]$ tuned-adm recommend
virtual-guest
[student@servera ~]$ sudo tuned-adm profile throughput-performance
[student@servera ~]$ tuned-adm active
Current active profile: throughput-performance
[student@servera ~]$ sudo tuned-adm profile "$(tuned-adm recommend)"
[student@servera ~]$ tuned-adm active
Current active profile: virtual-guest

Remember to go back to the recommended profile unless you measured a benefit. (Package installation is covered in chapter 13.)

Grade and finish

Check your work with lab grade sa-monitoring, then put the exercise away with lab finish sa-monitoring.