1
0
Fork 0
Tuning Profile Delivery Mechanism for Linux
Find a file
Jaroslav Škarvada 613169bc73 functions: used sysfs for governor switching by default
Patching of cpuspeed kept for backward compatibility.

Related: rhbz#1313489

Signed-off-by: Jaroslav Škarvada <jskarvad@redhat.com>
2016-11-29 16:58:39 +01:00
contrib {disk,net}devstat: fix typo in usage 2012-12-07 14:06:09 +01:00
doc man: fixed typo in tuned-profiles-oracle 2015-11-24 17:25:47 +01:00
ktune ktune: do not revert sysctl settings on ktune stop 2015-11-24 17:44:23 +01:00
libexec pmqos-static: fixed insecure pid file permissions 2013-02-27 18:00:44 +01:00
monitorplugins net: ignore uknown interfaces (e.g. bridges) 2016-04-15 12:12:13 +02:00
tests profiles: reversed SYSCTL_POST="/etc/sysctl.d/* /etc/sysctl.conf" 2015-02-09 14:22:21 +01:00
tune-profiles functions: used sysfs for governor switching by default 2016-11-29 16:58:39 +01:00
tuningplugins cpu plugin: add EEEPC dynamic FSB reclocking 2011-03-14 19:29:30 +01:00
udev udev rules: ask tuned to restart ktune instead of restarting it directly 2013-01-29 16:13:18 +01:00
.gitignore Specified commands for tuned-adm (default, server, off). 2009-07-27 16:36:24 +02:00
AUTHORS Integrated ktune-0.4 2009-07-27 17:24:58 +02:00
ChangeLog new release (0.2.23) 2013-01-29 16:13:33 +01:00
COPYING - Added copyright notices to files 2009-02-02 14:55:24 +01:00
git-daemon-export-ok - Initial stuff 2009-02-02 12:33:41 +01:00
INSTALL - Lots of whitespace changes 2009-02-25 19:22:54 +01:00
Makefile man: oracle profile documented in individual manual page 2015-11-24 15:44:13 +01:00
NEWS - Lots of updates to packaging and installation (specfile, makefile) 2009-02-20 18:29:35 +01:00
README - Large update to documentation and manpages 2010-03-19 17:38:41 +01:00
README.command Specified commands for tuned-adm (default, server, off). 2009-07-27 16:36:24 +02:00
tuned tuned: restart ktune 10 seconds after last SIGUSR2 was received 2013-01-29 16:09:23 +01:00
tuned-adm tuned-adm: fixed typo in the built-in help 2013-07-24 14:06:07 +02:00
tuned-adm.consolehelper allow run tuned-adm as root for users at the physical console 2009-10-16 14:16:28 +02:00
tuned-adm.pam allow run tuned-adm as root for users at the physical console 2009-10-16 14:16:28 +02:00
tuned.bash bash completion for tuned and tuned-adm 2010-06-01 19:03:04 +02:00
tuned.conf tuned: pid file moved to /var/run/tuned/tuned.pid 2010-07-13 15:50:24 +02:00
tuned.initscript initiscript: don't start tuned daemon on s390/s390x 2011-01-04 17:41:20 +01:00
tuned.py tuned: create PID file with proper permissions 2012-09-24 15:08:19 +02:00
tuned.service systemd: service file for tuned daemon 2011-07-20 17:10:23 +02:00
tuned.spec man: oracle profile documented in individual manual page 2015-11-24 15:44:13 +01:00
tuned.tmpfiles tmpfiles.d config to autocreate runtime directory on F15 2010-11-29 11:37:21 +01:00
tuned_adm.py tuned-adm: sort list of available profiles 2012-01-11 13:38:28 +01:00
tuned_logging.py logging: handlers initialization by need 2010-04-12 14:40:06 +02:00
tuned_nettool.py fix: crash when ethtool advertised modes in unknown format 2011-03-23 14:56:26 +01:00

tuned: A simple daemon that performs monitoring and adaptive configuration
       of devices in the system

Copyright (C) 2008, 2009 Red Hat, Inc.
Authors: Phil Knirsch

This program is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License
as published by the Free Software Foundation; either version 2
of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.


How to use
----------

After installing the package you simply need to start the tuned
service either with system-config-services or from the
commandline in a rootshell via:

  service tuned start

You can also automatically start it during system startup by
switching it on via chkconfig:

  chkconfig tuned on

After that the daemon will run automatically and monitor the
usage of various system components. Based on that information
components will then be put into lower or higher power saving
modes to adapt to the current usage. Currently plugins for CPU,
ethernet network and and ATA harddisk devices are implemented.

The config file /etc/tuned.conf contains several sections in
INI style. There is always a "main" section which controls
overall options of the daemon. Currently only 1 option
is available:

interval:  Interval for monitoring and tuning in seconds.
           Default is 10.

Each monitoring and tuning plugin has an own section then.
Currently only 1 option for plugins is available:

enabled: Enabled or disable the plugin. Default is True.
         Any other value disables it.


Monitoring/tuning plugins in detail
-----------------------------------

 - CPU
   This plugin pair monitors the CPU usage of the system. If the current
   workload of it is very low (uptime smaller than 0.2) it will switch
   the system wide PM-QOS setting to 999 usec. If the load should get
   higher again the PM-QOS latency will be lowered to 100 again. More details
   about PM-QOS can be found here http://www.lesswatts.org/projects/power-qos/

 - Disk
   This plugin pair monitors usage of every disk in the system. For each
   harddisk we check how long they have been in idle mode. The longer they are
   the more we raise the power saving features for that drive and allow it to
   actually spin down. As soon as it is being used we then reduce the power
   savings for it and deactivate the automatic spin down again at higher
   performance levels again.

 - Net
   This plugin pair monitors usage of network devices. At startup it checks
   what speed the interfaces have. If at any time the utilization of any
   interface falls below 5% of the possible maximum speed we then renegotiate
   the speed of the interface at 100mbit. In case utilization goes up again
   we renegotiate to the full speed once more.