Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
merge of bootchart-collector and pybootchartgui
Python C

This branch is 6 commits ahead, 340 commits behind master

Fetching latest commit…

Cannot retrieve the latest commit at this time

Failed to load latest commit information.
collector
examples
pybootchartgui
.gitignore
AUTHORS
COPYING
Makefile
NEWS
README
README.pybootchart
TODO
a.c
bootchartd
bootchartd.conf
pybootchartgui.py
setup.py

README

			    bootchart2
			   ------------

	bootchart2 was created from the fusion of three separate
pieces of work. First - the original bootchart: a shell script, and a
Java visualisation tool written by Ziga Mahkovec. Some of the original
shell scripting, and the concept remain unchanged from this time.

	bootchart2 replaces the Java visualisation with the more
friendly and flexible pybootchartgui (cf. README.pybootchart) written
by Anders Norgaard and Henning Niss, this lives mostly in the
pybootchart/ sub-directory.

	bootchart2 embeds a new collector, based on a port to C of the
inner-loop of the original bootchart collector shell-script by Scott
James Remnant. This has been subsequently re-written by Michael Meeks
to use the higher granularity 'taskstat' data available via a twisted
netlink interface.


			    Using bootchart2 ?
			   --------------------

	After install, simply add these options to your kernel
command-line, normally in /boot/grub/menu.lst:

	initcall_debug printk.time=y quiet init=/sbin/bootchartd ...

	Then - after bootup, run 'pybootchartgui -i' to get an interactive
chart rendering tool.


			    Why bootchart2 ?
			   ------------------

	There are a number of interesting additional features:

	* higher resolution - the taskstat interface gives nanosecond
	  timing information, where /proc/*/stat information is far,
	  far less reliable and useful.

	* higher performance - the C re-write allows us to collect more
	  data, more quickly - sampling at ~50+Hz.

	* no Java dependency - with the visualisation in easy-to-hack
	  python, development is quicker, and dependencies more commonly
	  found.

	* built-in visualisation - allowing some level of interaction,
	  zoom and so on.

	* better rendering - render to png, or svg, with added event
	  annotation support.

	* cumulative, and total CPU time graphing in the boot-chart.

	* and no doubt more by the time this is actually read.


			    Contributing
			   --------------

	For potential future work, see the TODO, and/or mail the authors,
all patches welcome.
Something went wrong with that request. Please try again.