Arch: AUR/minq-ananicy-git.
paru -S minq-ananicy-gitoryay -S minq-ananicy-gitshould do the trick. Don't forget to start and enable the service by runningsudo systemctl start ananicy.serviceandsudo systemctl enable ananicy.service.
For more information click here
More, updated and better organised rules.
Support for cmdline in rules added. This is particularly useful for applications that share the same name (looking at you Java). See the freenet rule as an example:
{ "name": "java", "cmdlines": ["freenet.node.NodeStarter"], "type": "service" }
This translates to: apply the rule to any process named java, that received freenet.node.NodeStarter as a command line argument. You can add more than one "cmdlines" in case you want to fine tune your rules.
Support for matching cmd by regexp:
{ "name": "ghc-.*", "name_is_regexp": true, "type": "compiler" }
Ananicy (ANother Auto NICe daemon) — is a shell daemon created to manage processes' IO and CPU priorities, with community-driven set of rules for popular applications (anyone may add their own rule via github's pull request mechanism). It's mainly for desktop usage.
I just wanted a tool for auto set programs nice in my system, i.e.:
- Why do I get lag, while compiling kernel and playing games?
- Why does dropbox client eat all my IO?
- Why does torrent/dc client make my laptop run slower?
- ...
Use ananicy to fix these problems!
X.Y.Z where
X - Major version,
Y - Script version - reset on each major update
Z - Rules version - reset on each script update
Read more about semantic versioning here
To use ananicy you must have systemd installed.
You can install ananicy manually by:
$ git clone https://github.com/Nefelim4ag/Ananicy.git /tmp/ananicy
$ cd /tmp/ananicy
$ sudo make install
- Debian/Ubuntu: use package.sh in repo
$ git clone https://github.com/kuche1/miinq-ananicy.git
$ ./minq-ananicy/package.sh debian
$ sudo dpkg -i ./minq-ananicy/ananicy-*.deb
Enable
$ sudo systemctl enable ananicy
$ sudo systemctl start ananicy
Rules files should be placed under /etc/ananicy.d/ directory and have *.rules extension.
Inside .rules file every process is described on a separate line. General syntax is described below:
{ "name": "gcc", "type": "Heavy_CPU", "nice": 19, "ioclass": "best-effort", "ionice": 7, "cgroup": "cpu90" }
All fields except name are optional.
name used for match processes by exec bin name
~ basename $(sudo realpath /proc/1/exe)
systemd
Currently matching by other things is not supported.
You can check what Ananicy sees, by:
ananicy dump proc
Ananicy loads all rules in ram while starting, so to apply rules, you must restart the service.
Available ionice values:
$ man ionice
CFQ IO Scheduller also uses nice for internal scheduling, so it's mean processes with same IO class and IO priority, but with different nicceness will take advantages of nice also for IO.
- Avoid changing
niceof system wide process like initrd. - Please try to use full process name (or name with
^$symbols likeNAME=^full_name$) - When writing rule - try to only use
nice, it must be enough in most cases. - Don't try set to high priority! Niceness can fix some performance problems, but can't give you more.
Example: pulseaudio uses
nice-11 by default, if you set other cpu hungry task, withnice{-20..-12} you can catch a sound glitches. - For CPU hungry backround task like compiling, just use
NICE=19.
About IO priority:
- It's useful to use
{"ioclass": "idle"}for IO hungry background tasks like: file indexers, Cloud Clients, Backups and etc. - It's not cool to set realtime to all tasks. The RT scheduling class is given first access to the disk, regardless of what else is going on in the system. Thus the RT class needs to be used with some care, as it can starve other processes. So try to use ioclass first.
Get ananicy output with journalctl:
$ journalctl -efu ananicy.service
If you see this error in the output
Jan 24 09:44:18 tony-dev ananicy[13783]: ERRO: Missing schedtool! Abort!
Fix it in Ubuntu with
sudo apt install schedtool
Please use pull request, thanks