Tune system performance
Linux #redhat #processes #daemons
Linux system comes fined tunned by default when you install, however there are a few tweaks that can be done based on
system performance and application requirements
In this note:
Optimize system performance by selecting a tuning profile managed by the tuned daemon
Prioritize or de-prioritize specific processes with the nice and renice commands
Optimize system performance
What is tuned ?
Tuned pronounced as tune-d
Tune is for system tuning and d is for daemon
It is systemd service that is used to tune Linux system performance
It is installed in CentOS/Redhat version 7, 8, and 9 by default.
tuned package name is tuned
The tuned service comes with pre-defined profiles and settings.
Based on selected profile the tuned service automictically adjust system to get the best performance. E.g. tuned will
adjust networking if you are downloading a large file or it will adjust IO settings if it detects high storage read/write
The tuned daemon applies system settings when the service starts or upon selection of a new tuning profile.
tuned profiles
Tuned profile Purpose
balanced deal for systems that require a compromise between power saving and performance
desktop Derived from the balanced profile. Provides faster response of interactive applications
Throughput- Tunes the system for maximum throughput
performance
Latency-performance Ideal for server systems that require low latency at the expense of power consumption
network-latency Derived from the latency-performance profile. It enables additional network tuning parameters to
provide low network latency
Network-throughput Derived from the throughput-performance profile. Additional network tuning parameters are applied
for maximum network throughput
powersave Tunes the system for maximum power saving
oracle Optimized for Oracle database loads based on the throughput-performance profile
virtual-guest Tunes the system for maximum performance if it runs on a virtual machine
virtual-host Tunes the system for maximum performance if it acts as a host for virtual machines
These are some tuned profiles, there could be added more but depending on the application.
Check if tuned is installed
You can run rpm -qa | grep tuned
Returned tuned-2.22.1-1.el9.noarch
Install tuned package if NOT installed already
dnf install tuned
Check tuned service status
systemctl status|enable|start tuned
When started, it will ask you to authenticate your user account.
Command to change setting for tuned daemon
tuned-adm
Command used to set the profile based on your system requirement
To check which profile is active
tuned-adm active
Example using tuned-adm command:
[user@localhost ~]$ tuned-adm active
Output:
Current active profile: virtual-guest
By default it has selected this profile, and the system will change tuned settings based on the profile it has selected.
To list all available profiles
tuned-adm list
Example using tuned-adm command:
[user@localhost ~]$ tuned-adm list
Output:
Available profiles:
- accelerator-performance - Throughput performance based tuning with disabled higher latency STOP
states
- aws - Optimize for aws ec2 instances
- balanced - General non-specialized tuned profile
- desktop - Optimize for the desktop use-case
- hpc-compute - Optimize for HPC compute workloads
- intel-sst - Configure for Intel Speed Select Base Frequency
- latency-performance - Optimize for deterministic performance at the cost of increased power
consumption
- network-latency - Optimize for deterministic performance at the cost of increased power
consumption, focused on low latency network performance
- network-throughput - Optimize for streaming network throughput, generally only necessary on
older CPUs or 40G+ networks
- optimize-serial-console - Optimize for serial console use.
- powersave - Optimize for low power consumption
- throughput-performance - Broadly applicable tuning that provides excellent performance across a
variety of common server workloads
- virtual-guest - Optimize for running inside a virtual guest
- virtual-host - Optimize for running KVM guests
Current active profile: virtual-guest
To change to desired profile
tuned-adm profile [profile-name]
Example using tuned-adm command:
[root@localhost ~]# tuned-adm profile balanced
Note we need to be root user in order to change the tuned profile.
In this line we are attempting to change to the balanced tuned profile.
Check for tuned recommendation
tuned-adm recommend
It will tell you what profile you should select based on the system that you are running.
Example using tuned-adm command:
[root@localhost ~]# tuned-adm recommend
Output:
virtual-guest
As we are running the system in a virtual machine, this is the best option we can select as profile.
Turn off tuned setting daemon
tuned-adm off
If you don't want the system to select any profile, you want to do things manually, you can turn tuned off.
Example using tuned-adm command:
[root@localhost ~]# tuned-adm active
Output:
No current active profile.
If the tuned-adm off command was ran previously, when trying to get the current profile, it will output this.
Change profile through web console
Login to https://192.168.1.x:9090
Remember to start the cockpit service, as well as stopping the firewalld .
Overview -> Configuration -> Performance profile
Then select the profile you want to set, and it will be applied right away.
Prioritize or de-prioritize processes.
Another way of keeping your system fine-tuned is by prioritizing processes through nice and renice command
If a server has 1 CPU then it can execute 1 computation/process at a time as they come in (first come first served) while
other processes must wait.
With nice and renice commands we can make the system to give preference to certain processes than others
This priority can be set at 40 different levels
The nice level values range from -20 (highest priority) to 19 (lowest priority) and by default, processes inherit their nice
level from their parent, which is usually 0.
The negative sign gives you the highest priority
To check process priority
top
Output from top :
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
824 rtkit 21 1 154108 3328 3072 S 0.3 0.0 0:02.63 rtkit-daemon
3835 root 20 0 0 0 0 I 0.3 0.0 0:03.89 kworker/u16:2-events_unbound
1 root 20 0 174564 17812 10756 S 0.0 0.2 0:15.90 systemd
2 root 20 0 0 0 0 S 0.0 0.0 0:00.31 kthreadd
When you run top , you will see two columns named PR and NI. (3rd and 4th column respectively)
The PR column is the system priority level
The NI is the actual user setting, the nice level that you actually set
What is the difference?
The system has its own priority that actually aligns it to the user nice level.
Nice value is a user-space and priority PR is the process's actual priority that use by Linux kernel. In Linux system
priorities are 0 to 139 in which 0 to 99 for real time and 100 to 139 for users
Process priority can be viewed through ps command as well with the right options
ps axo pid,comm,nice,cls --sort=-nice
Example using ps command:
[root@localhost ~]# ps axo pid,comm,nice,cls --sort=-nice
Output:
PID COMMAND NI CLS
80 khugepaged 19 TS
846 alsactl - IDL
79 ksmd 5 TS
824 rtkit-daemon 1 TS
1 systemd 0 TS
2 kthreadd 0 TS
12 rcu_tasks_kthre 0 TS
...
You'll see all the processes that are running in your system with priority level assigned to them.
The nice and renice commands
To set the process priority
nice -n [#] [process-name]
e.g. nice -n -15 top
Example using nice command:
[root@localhost ~]# nice -n -15 top
Changing the nice level of the top process to -15
The -n options adds integer N to the niceness (default is 10)
Output from top (it was ran a second time):
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
12173 root 5 -15 226056 4352 3456 R 0.3 0.0 0:00.03 top
Note that the new PR value is 5 and the new nice value is -15 (set previously)
To change the process priority
What if you wanted to change the process priority but without stopping the process. Use the renice command.
renice -n [#] [PID]
e.g. renice -n 12 12173
In this case you cannot use the process name, you have to use the process ID (PID).
Example using renice command:
[root@localhost ~]# renice -n 12 12173
Number 12173 is the PID of the top process.
We are attempting to change the process priority from the top process to a priority of 12, without having to stop and run
the process again.
Output:
12173 (process ID) old priority -15, new priority 12
Output from top (never stopped):
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
12173 root 32 12 226056 4352 3456 R 0.3 0.0 0:00.03 top
