duc - index, query and graph disk usage
duc subcommand options
Duc is a collection of tools for inspecting and visualizing disk usage.
Duc maintains an indexed database of accumulated sizes of directories of your file system, and allows you to query this database with some tools, or create fancy sunburst graphs to show you where your bytes are.
Duc scales quite well, it has been tested on systems with more than 500 million files and several petabytes of storage.
Duc comes with a command line tool called duc, which is used to create,
maintain and query the disk usage database. run duc help to get a list of
available commands. duc help <subcommand> describes the usage of a specific
subcommand. Run duc help --all for an extensive list of all commands and
their options.
Some commands might not be available on your system, depending on the exact
configuration chosen when building Duc. (For example, the duc gui command is
not available in the duc-nox package on Debian and Ubuntu)
Duc allows any option to be placed either on the command line or in a
configuration file. Options on the command line are preceded by a
double-leading-dash (--option), some options have a corresponding short
option which can be used as well with a single leading dash. (-o)
At startup duc tries to read its configuration from three locations in this
particular order: /etc/ducrc, ~/.config/duc/ducrc, ~/.ducrc and
./.ducrc.
A configuration file consists of sections and parameters. The section names correspond to the duc subcommands for which the parameters in that section apply. A section begins with the name of the section in square brackets and continues until the next section begins. Sections contain parameters, one per line, which consist of a single option name for boolean flags, or an option name and a value for options which take a value. See the EXAMPLES section for an example of the configuration file format.
Duc needs an index file of the file system before it is able to show any
information. To create the index, run the duc index command. For example, to
create an index of your home directory run duc index ~
$ duc index /usr
Skipping lost+found: Permission denied
Indexed 333823 files and 48200 directories, (35.0GB total) in 1 seconds
The default location of the database is $HOME/.duc.db. To use a different
database location, use the DUC_DATABASE environment variable or specify the
database location with the --database argument.
You can run duc index at any time later to rebuild the index.
By default Duc indexes all directories it encounters during file system traversal, including special file systems like /proc and /sys, and network file systems like NFS or Samba mounts. There are a few options to select what parts of your filesystem you want to include or exclude from the scan, check the documentation below for the options --one-file-system, --exclude, --fs-exclude and --fs-include for more details.
Duc has various subcommands for querying or exploring the index: (Note that depending on your configuration, some of these commands might not be available)
duc info shows a list of available directory trees in the database, and the time
and date of the last scan.
duc ls lists all files and directories under the given path on the console.
duc ui runs a ncurses based console user interface for exploring the file
system usage.
duc gui starts a graphical (X11) interface representing the file system in
a sunburst graph. Click on a directory to redraw the graph from the
perspective of the selected directory. Click in the center of the graph to go
up one directory in the tree.
This section list all available subcommands and describes their usage and options.
These options apply to all Duc subcommands:
--debug-h, --help
-q, --quiet
-v, --verbose
--versionOptions for command duc help [options]:
-a, --all
The 'index' subcommand performs a recursive scan of the given paths on the filesystem and calculates the inclusive size of all directories. The results are written to the index, and can later be queried by one of the other duc tools.
Options for command duc index [options] PATH ...:
-b, --bytes
-d, --database=VAL
-e, --exclude=VAL
-H, --check-hard-links
-f, --force
--fs-exclude=VAL--fs-include=VAL--hide-file-names-U, --uid=VAL
-u, --username=VAL
-m, --max-depth=VAL
-x, --one-file-system
-p, --progress
--dry-run--uncompressedOptions for command duc info [options]:
-a, --apparent
-b, --bytes
-d, --database=VAL
The 'ls' subcommand queries the duc database and lists the inclusive size of all files and directories on the given path. If no path is given the current working directory is listed.
Options for command duc ls [options] [PATH]...:
-a, --apparent
--ascii-b, --bytes
-F, --classify
-c, --color
--count-d, --database=VAL
-D, --directory
--dirs-only--full-path-g, --graph
-l, --levels=VAL
-n, --name-sort
-R, --recursive
Options for command duc xml [options] [PATH]:
-a, --apparent
-d, --database=VAL
-x, --exclude-files
-s, --min_size=VAL
Options for command duc json [options] [PATH]:
-a, --apparent
-d, --database=VAL
-x, --exclude-files
-s, --min_size=VAL
The 'graph' subcommand queries the duc database and generates a sunburst graph showing the disk usage of the given path. If no path is given a graph is created for the current working directory.
By default the graph is written to the file 'duc.png', which can be overridden by using the -o/--output option. The output can be sent to stdout by using the special file name '-'.
Options for command duc graph [options] [PATH]:
-a, --apparent
-d, --database=VAL
--count--dpi=VAL-f, --format=VAL
--fuzz=VAL--gradient-l, --levels=VAL
-o, --output=VAL
--palette=VAL--ring-gap=VAL-s, --size=VAL
Options for command duc cgi [options] [PATH]:
-a, --apparent
-b, --bytes
--count--css-url=VAL-d, --database=VAL
--dpi=VAL--footer=VAL--fuzz=VAL--gradient--header=VAL-l, --levels=VAL
--list--palette=VAL--ring-gap=VAL-s, --size=VAL
--tooltipThe 'gui' subcommand queries the duc database and runs an interactive graphical utility for exploring the disk usage of the given path. If no path is given the current working directory is explored.
The following keys can be used to navigate and alter the graph:
+ increase maximum graph depth
- decrease maximum graph depth
0 Set default graph depth
a Toggle between apparent and actual disk usage
b Toggle between exact byte count and abbreviated sizes
c Toggle between file size and file count
f toggle graph fuzz
g toggle graph gradient
p toggle palettes
backspace go up one directory
Options for command duc gui [options] [PATH]:
-a, --apparent
-b, --bytes
--count--dark-d, --database=VAL
--fuzz=VAL--gradient-l, --levels=VAL
--palette=VAL--ring-gap=VALThe 'ui' subcommand queries the duc database and runs an interactive ncurses utility for exploring the disk usage of the given path. If no path is given the current working directory is explored.
The following keys can be used to navigate and alter the file system:
up, pgup, j: move cursor up
down, pgdn, k: move cursor down
home, 0: move cursor to top
end, $: move cursor to bottom
left, backspace: go up to parent directory (..)
right, enter: descent into selected directory
a: toggle between actual and apparent disk usage
b: toggle between exact and abbreviated sizes
c: Toggle between file size and file count
h: show help. press 'q' to return to the main screen
n: toggle sort order between 'size' and 'name'
o: try to open the file using xdg-open
q, escape: quit
Options for command duc ui [options] [PATH]:
-a, --apparent
-b, --bytes
--count-d, --database=VAL
-n, --name-sort
--no-colorThe duc binary has support for a rudimentary CGI interface, currently only
tested with apache. The CGI interface generates a simple HTML page with a list
of indexed directories, and shows a clickable graph for navigating the file
system. If the option --list is given, a list of top sized files/dirs is also
written.
Configuration is done by creating a simple shell script as .cgi in a directory
which is configured for CGI execution by your web server (usually
/usr/lib/cgi-bin). The shell script should simply start duc, and pass the
location of the database to navigate.
An example duc.cgi script would be
#!/bin/sh
/usr/local/bin/duc cgi -d /home/jenny/.duc.db
chmod +x duc.cgi)Some notes:
The HTML page is generated with a simple embedded CSS style sheet. If the style is not to your liking you can provide an external CSS url with the --css-url option which will then be used instead of the embedded style definition.
Add the option --list to generate a table of top sized files and directories in the HTML page.
The options --header and --footer allow you to insert your own HTML code before and after the main.
The current CGI configuration is not very flexible, nor secure. It is not advised to run the CGI from public reachable web servers, use at your own risk.
The concepts of 'file size' and 'disk usage' can be a bit confusing. Files on
disk have an apparent size, which indicates how much bytes are in the file from
the users point of view; this is the size reported by tools like ls -l. The
apparent size can be any number, from 0 bytes up to several TB. The actual
number of bytes which are used on the filesystem to store the file can differ
from this apparent size for a number of reasons: disks store data in blocks,
which cause files to always take up a multiple of the block size, files can
have holes ('sparse' files), and other technical reasons. This number is always
a multiple of 512, which means that the actual size used for a file is almost
always a bit more than its apparent size.
Duc has two modes for counting file sizes:
apparent size: this is the size as reported by ls. This number indicates
the file length, which is usually smaller than the actual disk usage.
actual size: this is the size as reported by du and df. The actual file
size tells you how much disk is actually used by a file, and is always a
multiple of 512 bytes.
The default mode used by duc is to use the 'actual size'. Most duc commands to
report disk usage (duc ls, duc graph, duc ui, etc) have an option to
change between these two modes (usually the -a), or use the 'a' key to
toggle.
If you use git clone to pull down the latest release, you will have to do the following:
git clone https://github.com/zevv/duc
cd duc
autoreconf -i
Then you can run the regular
./configure [ options ]
make
to the regular build of the software.
A note for Redhat and derivates users. The package providing the development file for lmdb (lmdb-devel) does not include a lmdb.pc pkgconfig file. This could lead to errors during the configure phase:
checking for LMDB... no
configure: error: Package requirements (lmdb) were not met:
To avoid the need to call pkg-config, you may set the environment variables
LMDB_CFLAGS and LMDB_LIBS:
LMDB_CFLAGS=" " LMDB_LIBS=-llmdb ./configure --with-db-backend=lmdb [ options ]
Index the /usr directory, writing to the default database location ~/.duc.db:
$ duc index /usr
List all files and directories under /usr/local, showing relative file sizes in a graph:
$ duc ls -Fg /usr/local
4.7G lib/ [+++++++++++++++++++++++++++++++++++++++++++]
3.1G share/ [++++++++++++++++++++++++++++ ]
2.7G src/ [++++++++++++++++++++++++ ]
814.9M bin/ [+++++++ ]
196.6M include/ [+ ]
66.6M x86_64-w64-mingw32/ [ ]
59.9M local/ [ ]
38.8M i686-w64-mingw32/ [ ]
20.3M sbin/ [ ]
13.6M lib32/ [ ]
13.3M libx32/ [ ]
or use the -R options for the tree view:
$ duc ls -RF /etc/logcheck
24.0K `+- ignore.d.server/
4.0K | `+- hddtemp
4.0K | |- ntpdate
4.0K | |- lirc
4.0K | |- rsyslog
4.0K | `- libsasl2-modules
8.0K |- ignore.d.workstation/
4.0K | `- lirc
8.0K `- ignore.d.paranoid/
4.0K `- lirc
Start the graphical interface to explore the file system using sunburst graphs:
$ duc gui /usr
Generate a graph of /usr/local in .png format:
$ duc graph -o /tmp/usr.png /usr
The following sample configuration file defines default parameters for the duc
ls and duc ui commands and defines a global option to configure the database
path which is used by all subcommands
[global]
database /var/cache/duc.db
[ls]
recursive
classify
color
[ui]
no-color
apparent
What does the error 'Database version mismatch mean?'
The layout of the index database sometimes changes when new features are implemented. When you get this error you have probably upgraded to a newer version. Just remove the old database file and rebuild the index.
Duc crashes with a segmentation fault, is it that buggy?
By default Duc uses the Tokyocabinet database backend. Tokyocabinet is pretty fast, stores the database in a single file and has nice compression support to keep the database small. Unfortunately, it is not always robust and sometimes chokes on corrupt database files. Try to remove the database and rebuild the index. If the error persists contact the authors.
Some of the Duc subcommands like duc gui are not available on my system?
Depending on the configuration that was chosen when building Duc, some
options might or might not be available in the duc utility. For example, on
Debian or Ubuntu Duc comes in two flavours: there is a full featured package
called duc, or a package without dependencies on X-windows called
duc-nox, for which the latter lacks the duc gui command.
duc index is hogging my system and using a lot of CPU and I/O!
Traversing a file system is hard work - which is the exact reason why Duc exists in the first place. You can use the default tools to make Duc behave nice towards other processes on your machine, use something like:
nice 19 ionice -c 3 duc index [options]
This makes duc index run with the lowest CPU and I/O scheduler priorities,
which is nicer to all the other processes on your machine.
At startup duc tries to read its configuration from three locations in this
particular order: /etc/ducrc, ~/.config/duc/ducrc, ~/.ducrc and
./.ducrc.
Duc mainains an index of scanned directories, which defaults to ~/.duc.db. All tools take the -d/--database option to override the database path.
Other contributors can be found in the Git log at GitHub.
Duc is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; version 3 dated June, 2007. Duc 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 Lesser General Public License for more details.