/usr/share/doc/reprepro/manual.html

<html><head>
<title>reprepro manual</title>
<!-- some style elements stolen from or inspired by bugs.debian.org /-->
<style>
<!--
html { color: #000; background: #fefefe; font-family: serif; margin: 1em; border: 0; padding: 0; line-height: 120%; }
body { margin: 0; border: 0; padding: 0; }
pre { text-align: left; border: #f0f0f0 1px solid;  padding: 1px;}
pre.shell { text-align: left; border: none; border-left: #f0f0f0 1px solid;  padding: 2px;}
h1, h2, h3 { text-align: left; font-family: sans-serif; background-color: #f0f0ff; color: #3c3c3c; border: #a7a7a7 1px solid;  padding: 10px;}
h1 { font-size: 180%; line-height: 150%; }
h2 { font-size: 150% }
h3 { font-size: 100% }
ul.dir { list-style-type: disc; }
ul { list-style-type: square; }
dt.dir, dt.file, dt.symlink { font-weight:bold; font-family: sans-serif; }
/-->
</style>
</head>
<body>
<h1>reprepro manual</h1>
This manual documents reprepro, a tool to generate and administer
Debian package repositories.
<br>
Other useful resources:
<ul>
<li> the <a href="http://mirrorer.alioth.debian.org/">homepage</a> of  reprepro.</li>
<li> <a href="file://localhost/usr/share/doc/reprepro/">local directory</a> with documentation and examples, if you have reprepro installed.</li>
<li> the <a href="http://git.debian.org/?p=mirrorer/reprepro.git;a=blob_plain;f=docs/FAQ;hb=HEAD">Frequently Asked Questions</a></li>
</ul>
<h2>Table of contents</h2>
Sections of this document:
<ul>
<li><a href="#introduction">Introduction</a></li>
<li><a href="#firststeps">First steps</a></li>
<li><a href="#dirbasics">Repository basics</a></li>
<li><a href="#config">Config files</a></li>
<li><a href="#export">Generation of index files</a>
 <ul>
 <li><a href="#compression">Compression and file names</a></li>
 <li><a href="#signing">Signing</a></li>
 <li><a href="#contents">Contents files</a></li>
 <li><a href="#exporthook">Additional index files (like .diff)</a></li>
 </ul></li>
<li><a href="#localpackages">Local packages</a>
 <ul>
 <li><a href="#include">Including via command line</a></li>
 <li><a href="#incoming">Processing an incoming queue</a></li>
 </ul></li>
<li><a href="#mirroring">Mirroring</a></li>
<li><a href="#propagation">Propagation of packages</a></li>
<li><a href="#snapshosts">Snapshots</a> (TODO)</li>
<li><a href="#tracking">Source package tracking</a> (TODO)</li>
<li><a href="#hooks">Extending reprepro / Hooks and more</a></li>
<li><a href="#maintenance">Maintenance</a></li>
<li><a href="#internals">Internals</a></li>
<li><a href="#recovery">Disaster recovery</a></li>
<li><a href="#paranoia">Paranoia</a></li>
<li><a href="#counterindications">What reprepro cannot do</a></li>
</ul>
<h2><a name="introduction">Introduction</a></h2>
<h3>What reprepro does</h3>
Reprepro is a tool to take care of a repository of Debian packages
(<tt>.dsc</tt>,<tt>.deb</tt> and <tt>.udeb</tt>).
It installs them to the proper places, generates indices of packages
(<tt>Packages</tt> and <tt>Sources</tt> and their compressed variants)
and of index files (<tt>Release</tt> and optionally <tt>Release.gpg</tt>),
so tools like <tt>apt</tt> know what is available and where to get it from.
It will keep track which file belongs to where and remove files no longer
needed (unless told to not do so).
It can also make (partial) partial mirrors of remote repositories,
including merging multiple sources and
automatically (if explicitly requested) removing packages no longer available
in the source.
And many other things (sometimes I fear it got a few features too much).
<h3>What reprepro needs</h3>
It needs some libraries (<tt>zlib</tt>, <tt>libgpgme</tt>, <tt>libdb</tt> (Version 3, 4.3 or 4.4)) and can be compiled with some more for additional features (<tt>libarchive</tt>,
<tt>libbz2</tt>).
Otherwise it only needs
<tt>apt</tt>'s methods (only when downloading stuff),
<tt>gpg</tt> (only when signing or checking signatures),
and if compiled without <tt>libarchive</tt> it needs <tt>tar</tt> and <tt>ar</tt> installed.
<br>
If you tell reprepro to call scripts for you, you will of course need the interpreters for these scripts:
The included example to generate pdiff files needs python. The example to extract
changelogs needs dpkg-source.
<h3>What this manual aims to do</h3>
This manual aims to give some overview over the most important features,
so people can use them and so that I do not implement something a second
time because I forgot support is already there.
For a full reference of all possible commands and config options take a
look at the man page, as this manual might miss some of the more obscure
options.
<h2><a name="firststeps">First steps</a></h2>
<h3>generate a repository with local packages</h3>
<ul>
<li>Choose a directory (or create it).</li>
<li>Create a subdirectory called <tt>conf</tt> in there.</li>
<li>In the <tt>conf/</tt> subdirectory create a file called <tt>distributions</tt>,
with content like:
<pre class="file">
Codename: mystuff
Components: main bad
Architectures: sparc i386 source
</pre>
or with content like:
<pre class="file">
Codename: andy
Suite: rusty
Components: main bad
Architectures: sparc i386 source
Origin: myorg
Version: 20.3
Description: my first little repository
</pre>
(Multiple distributions are separated by empty lines, Origin, Version and Description
are just copied to the generated Release files, more things controlling reprepro can
appear which are described later).
</li>
<li>If your <tt>conf/distributions</tt> file contained a <tt>Suite:</tt> and you
are too lazy to generate the symbolic links yourself, call:
<pre class="shell">
reprepro -b $YOURBASEDIR createsymlinks
</pre>
</li>
<li>Include some package, like:
<pre class="shell">
reprepro -b $YOURBASEDIR include mystuff mypackage.changes
</pre>
or:
<pre class="shell">
reprepro -b $YOURBASEDIR includedeb mystuff mypackage.deb
</pre>
</li>
<li>Take a look at the generated <tt>pool</tt> and <tt>dists</tt>
directories. They contain everything needed to apt-get from.
Tell apt to include it by adding the following to your <tt>sources.list</tt>:
<pre class="file">
deb file:///$YOURBASEDIR mystuff main bad
</pre>
or make it available via http or ftp and do the same <tt>http://</tt> or <tt>ftp://</tt> source.</li>
</ul>
<h3>mirroring packages from other repositories</h3>
This example shows how to generate a mirror of a single architecture with
all packages of etch plus security updates:
<ul>
<li>Choose a directory (or create it).</li>
<li>Create a subdirectory called <tt>conf</tt> in there (if not already existent).</li>
<li>In the <tt>conf/</tt> subdirectory create a file called <tt>distributions</tt>,
with content like (or add to that file after an empty line):
<pre class="file">
Origin: Debian
Label: Debian
Suite: stable
Version: 4.0
Codename: etch
Architectures: i386
Components: main
Description: Debian 4.0 etch + security updates
Update: - debian security
Log: logfile
</pre>
Actually only <tt>Codename</tt>, <tt>Components</tt>, <tt>Architecture</tt> and <tt>Update</tt> is needed, the rest is just information for clients.
The <tt>Update</tt> line tells to delete everything no longer available (<tt>-</tt>),
then add the <tt>debian</tt> and <tt>security</tt> rules, which still have to be defined:
</li>
<li>In the <tt>conf/</tt> subdirectory create a file called <tt>updates</tt>,
with content like (or add to that file after an empty line:):
or with content like:
<pre class="file">
Name: security
Method: http://security.debian.org/debian-security
Fallback: ftp://klecker.debian.org/debian-security
Suite: */updates
VerifyRelease: A70DAF536070D3A1|B5D0C804ADB11277
Architectures: i386
Components: main
UDebComponents:

Name: debian
Method: http://ftp2.de.debian.org/debian
Config: Acquire::Http::Proxy=http://proxy.myorg.de:8080
VerifyRelease: A70DAF536070D3A1|B5D0C804ADB11277
</pre>
(If there are no Architecture, Components or UDebComponents, it will try all the distribution to update has. Fallback means a URL to try when the first cannot offer some file (Has to be the same method)).
</li>
<li>Tell reprepro to update:
<pre class="shell">
reprepro -b $YOURBASEDIR update etch
</pre>
</li>
<li>Take a look at at the generated <tt>pool</tt> and <tt>dists</tt>
directories. They contain everything needed to apt-get from.
Tell apt to include it by adding the following to your <tt>sources.list</tt>:
<pre class="shell">
deb file:///$YOURBASEDIR etch main
</pre>
or make it available via http or ftp.</li>
</ul>
<h2><a name="dirbasics">Repository basics</a></h2>
An <tt>apt-get</tt>able repository of Debian packages consists of two parts:
the index files describing what is available and where it is and the actual
Debian binary (<tt class="suffix">.deb</tt>),
installer binary (<tt class="suffix">.udeb</tt>),
and source (<tt class="suffix">.dsc</tt> together with
<tt class="suffix">.tar.gz</tt> or
<tt class="suffix">.orig.tar.gz</tt> and
<tt class="suffix">.diff.gz</tt>) packages.
<br>
While you do not know how these look like to use reprepro, it's always a good
idea to know what you are creating.
<h3>Index files</h3>
All index files are in subdirectories of a directory called
<tt class="dirname">dists</tt>. Apt is very decided what names those should
have, including the name of <tt class="dirname">dists</tt>.
Including all optional and extensional files, the hierarchy looks like this:

<dl class="dir">
<dt class="dir">dists</dt><dd>
	<dl class="dir">
	<dt class="dir">CODENAME</dt><dd>
Each distribution has it's own subdirectory here, named by it's codename.
		<dl class="dir">
		<dt class="file">Release</dt><dd>
This file describes what distribution this is and the checksums of
all index files included.
		</dd>
		<dt class="file">Release.gpg</dt><dd>
This is the optional detached gpg signature of the Release file.
Take a look at the <a name="#signing">section about signing</a> for how to
active this.
		</dd>
		<dt class="file">Contents-ARCHITECTURE.gz</dt><dd>
This optional file lists all files and which packages they belong to.
It's downloaded and used by tools like
<a href="http://packages.debian.org/apt-file">apt-file</a>
to allow users to determine which package to install to get a specific file.
<br>
To activate generating of these files by reprepro, you need a <a href="#contents">Contents</a>
header in your distribution declaration.
		</dd>
		<dt class="dir">COMPONENT1</dt><dd>
Each component has it's own subdirectory here. They can be named whatever users
can be bothered to write into their <tt class="filename">sources.list</tt>, but
things like <tt>main</tt>, <tt>non-free</tt> and <tt>contrib</tt> are common.
But funny names like <tt>bad</tt> or <tt>universe</tt> are just as possible.
			<dl class="dir">
			<dt class="dir">source</dt><dd>
If this distribution supports sources, this directory lists which source
packages are available in this component.
				<dl class="dir">
				<dt class="file">Release</dt><dd>
This file contains a copy of those information about the distribution
applicable to this directory.
				</dd>
				<dt class="file">Sources</dt>
				<dt class="file">Sources.gz</dt>
				<dt class="file">Sources.bz2</dt><dd>
These files contain the actual description of the source Packages. By default
only the <tt class="suffix">.gz</tt> file created, to create all three add the
following to the declarations of the distributions:
<pre class="config">
DscIndices Sources Release . .gz .bz2
</pre>
That header can also be used to name those files differently, but then apt
will no longer find them...
				</dd>
				<dt class="dir">Sources.diff</dt><dd>
This optional directory contains diffs, so that only parts of the index
file must be downloaded if it changed. While reprepro cannot generate these
so-called <tt>pdiff</tt>s itself, it ships both with a program called rredtool
and with an example python script to generate those.
				</dd>
				</dl>
			</dd>
			</dl>
			<dl class="dir">
			<dt class="dir">binary-ARCHITECTURE</dt><dd>
Each architecture has its own directory in each component.
				<dl class="dir">
				<dt class="file">Release</dt><dd>
This file contains a copy of those information about the distribution
applicable to this directory.
				</dd>
				<dt class="file">Packages</dt>
				<dt class="file">Packages.gz</dt>
				<dt class="file">Packages.bz2</dt><dd>
These files contain the actual description of the binary Packages. By default
only the uncompressed and <tt class="suffix">.gz</tt> files are created.
To create all three, add the following to the declarations of the distributions:
<pre class="config">
DebIndices Packages Release . .gz .bz2
</pre>
That header can also be used to name those files differently, but then apt
will no longer find them...
				</dd>
				<dt class="dir">Packages.diff</dt><dd>
This optional directory contains diffs, so that only parts of the index
file must be downloaded if it changed. While reprepro cannot generate these
so-called <tt>pdiff</tt>s itself, it ships both with a program called rredtool
and with an example python script to generate those.
				</dd>
				</dl>
			</dd>
			<dt class="dir">debian-installer</dt><dd>
This directory contains information about the <tt class="suffix">.udeb</tt>
modules for the <a href="http://www.debian.org/devel/debian-installer/">Debian-Installer</a>.
Those are actually just a very stripped down form of normal <tt class="suffix">.deb</tt>
packages and this the hierarchy looks very similar:

				<dl class="dir">
				<dt class="dir">binary-ARCHITECTURE</dt><dd>
					<dl class="dir">
					<dt class="file">Packages</dt><dd></dd>
					<dt class="file">Packages.gz</dt><dd></dd>
					</dl>
				</dd>
				</dl>
			</dd>
			</dl>
		</dd>
		<dt class="dir">COMPONENT2</dt><dd>
There is one dir for every component. All look just the same.
		</dd>
		</dl>
	</dd>
	<dt class="symlink">SUITE -> CODENAME</dt><dd>
To allow accessing distribution by function instead of by name, there are often
symbolic links from suite to codenames. That way users can write
<pre class="config">
deb http://some.domain.tld/debian SUITE COMPONENT1 COMPONENT2
</pre>
instead of
<pre class="config">
deb http://some.domain.tld/debian CODENAME COMPONENT1 COMPONENT2
</pre>
in their <tt class="filename">/etc/apt/sources.list</tt> and totally get
surprised by getting something new after a release.
	</dd>
	</dl>
</dd></dl>
<h3>Package pool</h3>
While the index files have a required filename, the actual files
are given just as relative path to the base directory you specify
in your sources list. That means apt can get them no matter what
scheme is used to place them. The classical way Debian used till
woody was to just put them in subdirectories of the
<tt class="dir">binary-ARCHITECTURE</tt> directories, with the exception
of the architecture-independent packages, which were put into a
artificial <tt class="dir">binary-all</tt> directory. This was replaced
for the official repository with package pools, which reprepro also uses.
(Actually reprepro stores everything in pool a bit longer than the official
repositories, that's why it recalculates all filenames without exception).
<br>
In a package pool, all package files of all distributions in that repository
are stored in a common directory hierarchy starting with <tt class="dir">pool/</tt>,
only separated by the component they belong to and the source package name.
As everything this has disadvantages and advantages:
<ul><li>disadvantages
	<ul><li>different files in different distributions must have different filenames
	</li><li>it's impossible to determine which distribution a file belongs to by path and filename (think mirroring)
	</li><li>packages can no longer be grouped together in common subdirectories by having similar functions
	</li></ul>
</li><li>advantages
	<ul><li>the extremely confusing situation of having differently build packages with the same version if different distributions gets impossible by design.
	</li><li>the source (well, if it exists) is in the same directory as the binaries generated from it
	</li><li>same files in different distributions need disk-space and bandwidth only once
	</li><li>each package can be found only knowing component and sourcename
	</li></ul>
</li></ul>
Now let's look at the actual structure of a pool (there is currently no difference
between the pool structure of official Debian repositories and those generated by
reprepro):

<dl class="dir">
<dt class="dir">pool</dt><dd>
	The directory all this resides is is normally called <tt class="dir">pool</tt>.
	That's nowhere hard coded in apt but that only looks at the relative
	directory names in the index files. But there is also no reason to name
	it differently.
	<dl class="dir">
	<dt class="dir">COMPONENT1</dt><dd>
Each component has it's own subdirectory here.
They can be named whatever users
can be bothered to write into their <tt class="filename">sources.list</tt>, but
things like <tt>main</tt>, <tt>non-free</tt> and <tt>contrib</tt> are common.
But funny names like <tt>bad</tt> or <tt>universe</tt> are just as possible.
		<dl class="dir">
		<dt class="dir">a</dt><dd>
As there are really many different source packages,
the directory would be too full when all put here.
So they are separated in different directories.
Source packages starting with <tt class="constant">lib</tt> are put into a
directory named after the first four letters of the source name.
Everything else is put in a directory having the first letter as name.
			<dl class="dir">
			<dt class="dir">asource</dt><dd>
Then the source package name follows.
So this directory <tt class="dir">pool/COMPONENT1/a/asource/</tt> would contain
all files of different versions of the hypothetical package <tt class="constant">asource</tt>.
				<dl class="dir">
				<dt class="dir">asource</dt><dd>
				<dt class="file">a-source_version.dsc</dt>
				<dt>a-source_version.tar.gz</dt><dd>
The actual source package consists of its description file (<tt class="suffix">.dsc</tt>)
and the files references by that.
				</dd>
				<dt class="file">binary_version_ARCH1deb</dt>
				<dt class="file">binary_version_ARCH2.deb</dt>
				<dt class="file">binary2_version_all.deb</dt><dd>
				<dt class="file">di-module_version_ARCH1.udeb</dt><dd>
Binary packages are stored here to.
So to know where a binary package is stored you need to know what its source package
name is.
				</dd>
				</dl>
			</dd>
			</dl>
		</dd>
		<dt class="dir">liba</dt><dd>
As described before packages starting with <tt class="constant">lib</tt> are not stored
in <tt class="dir">l</tt> but get a bit more context.
		</dd>
		</dl>
	</dd>
	<dt class="dir">COMPONENT2</dt><dd>
There is one dir for every component. All look just the same.
	</dd>
	</dl>
</dd></dl>
As said before, you don't need to know this hierarchy in normal operation.
reprepro will put everything to where it belong, keep account what is there
and needed by what distribution or snapshot, and delete files no longer needed.
(Unless told otherwise or when you are using the low-level commands).
<h2><a name="config">Config files</a></h2>
Configuring a reprepro repository is done by writing some config files
into a directory.
This directory is currently the <tt class="dir">conf</tt> subdirectory of the
base directory of the repository,
unless you specify <tt class="option">--confdir</tt> or set the
environment variable <tt class="env">REPREPRO_CONFIG_DIR</tt>.

<dl class="dir">
<dt class="dir">options</dt><dd>
If this file exists, reprepro will consider each line an additional
command line option.
Arguments must be in the same line after an equal sign.

Options specified on the command line take precedence.
</dd>
<dt class="dir">distributions</dt><dd>
This is the main configuration file and the only that is needed in all
cases.
It lists the distributions this repository contains and their properties.
<br>
See <a href="#firststeps">First steps</a> for a short example or the manpage
for a list of all possible fields.
</dd>
<dt class="dir">updates</dt><dd>
Rules about where to download packages from other repositories.
See the section <a href="#mirroring">Mirroring / Updating</a>
for more examples or the man page for a full reference.
</dd>
<dt class="dir">pulls</dt><dd>
Rules about how to move packages in bulk between distributions
where to download packages from other repositories.
See the section <a href="#propagation">Propagation of packages</a>
for an example or the man page for full reference.
</dd>
<dt class="dir">incoming</dt><dd>
Rules for incoming queues as processed by <tt class="command">processincoming</tt>.
See <a href="#processincoming-incoming-config">Processing an incoming queue</a> for more information.
</dd>
</dl>
<h2><a name="export">Generation of index files</a></h2>
<h3>Deciding when to generate</h3>
As reprepro stores all state in its database,
you can decide when you want them to be written to the <tt class="dir">dists/</tt>
directory.
You can always tell reprepro to generate those files with the <tt>export</tt> command:
<pre class="command">
reprepro -b $YOURBASEDIR export $CODENAMES
</pre>
This can be especially useful, if you just edited <tt class="file">conf/distributions</tt>
and want to test what it generates.
<p>
While that command regenerates all files, in normal operation reprepro will only
regenerate files where something just changed or that are missing.
With <tt class="option">--export</tt> option you can control when this fill happen:
<dl><dt>never</dt><dd>Don't touch any index files.
This can be useful for doing multiple operations in a row and not wanting to regenerate
the indices all the time.
Note that unless you do an explicit export or change the same parts later without that
option, the generated index files may be permanently out of date.
</dd><dt>changed</dt><dd>This is the default behaviour since 3.0.1.
Only export distributions where something changed
(and no error occoured that makes an inconsistent state likely).
And in those distributions only (re-)generate files which content should have been changed
by the current action or which are missing.
</dd><dt>lookedat</dt><dd>New name for <tt>normal</tt> since 3.0.1.
</dd><dt>normal</dt><dd>This was the default behaviour until 3.0.0 (changed in 3.0.1).
In this mode all distributions are processed that were looked at without error
(where error means only errors hapening while the package was open so have a chance
to cause strange contents).
This ensures that even after a operation that had nothing to do the looked at
distribution has all the files exported needed to access it. (But still only files
missing or that content would change with this action are regenerated).
</dd><dt>force</dt><dd>Also try to write the current state if some error occured.
In all other modes reprepro will not write the index files if there was a problem.
While this keeps the repository usable for users, it means that you will need an
explicit export to write possible other changes done before that in the same run.
(reprepro will tell you that at the end of the run with error, but you should not
miss it).
</dd></dl>
<h3>Distribution specific fields</h3>
There are a lot of <tt class="file">conf/distributions</tt> headers to control
what index files to generate for some distribution, how to name
them, how to postprocess them and so on. The most important are:
<h4>Fields for the Release files</h4>
The following headers are copied verbatim to the Release file, if they exist:
<tt class="header">Origin</tt>,
<tt class="header">Label</tt>,
<tt class="header">Codename</tt>,
<tt class="header">Suite</tt>,
<tt class="header">Architectures</tt> (excluding a possible value "<tt>source</tt>"),
<tt class="header">Components</tt>,
<tt class="header">Description</tt>, and
<tt class="header">NotAutomatic</tt>,
<tt class="header">ButAutomaticUpgrades</tt>.
<h4><a name="compression">Choosing compression and file names</a></h4>
Depending on the type of the index files, different files are generated.
No specifying anything is equivalent to:
<pre class="config">
 DscIndices Sources Release .gz
 DebIndices Packages Release . .gz
 UDebIndices Packages . .gz
</pre>
This means to generate <tt>Release</tt>, <tt>Sources.gz</tt> for sources,
<tt>Release</tt>, <tt>Packages</tt> and <tt>Packages.gz</tt> for binaries
and <tt>Packages</tt> and <tt>Packages.gz</tt> for installer modules.
<br>
The format of these headers is the name of index file to generate, followed
by the optional name for a per-directory release description
(when no name is specified, no file is generated).
Then a list of compressions:
A single dot (<tt>.</tt>) means generating an uncompressed index,
<tt>.gz</tt> means generating a gzipped output,
while <tt>.bz2</tt> requests and bzip2ed file.
(<tt>.bz2</tt> is not available when disabled at compile time).
After the compressions a script can be given that is called to generate/update
additional forms, see <a href="#exporthook">&quot;Additional index files&quot;</a>.
<h4><a name="signing">Signing</a></h4>
If there is a <tt class="config">SignWith</tt> header, reprepro will try
to generate a <tt class="file">Release.gpg</tt> file using libgpgme.
If the value of the header is <tt>yes</tt> it will use the first key
it finds, otherwise it will give the option to libgpgme to determine the
key. (Which means fingerprints and keyids work fine, and whatever libgpgme
supports, which might include most that gpg supports to select a key).
<br>
The best way to deal with keys needing passphrases is to use
<a href="http://packages.debian.org/gnupg-agent">gpg-agent</a>.
The only way to specify which keyring to use is to set the
<tt class="env">GNUPGHOME</tt> enviroment variable, which will effect all
distributions.
<h4><a name="contents">Contents files</a></h4>
Reprepro can generate files called
<tt class="file">dists/CODENAME/Contents-ARCHITECTURE.gz</tt>
listing all files in all binary packages available for the selected
architecture in that distribution and which package they belong to.
<br>
This file can either be used by humans directly or via downloaded
and searched with tools like
<a href="http://packages.debian.org/apt-file">apt-file</a>.
<br>
To activate generating of these files by reprepro, you need a <tt class="config">Contents</tt> header in that distribution's declaration in <tt class="file">conf/distributions</tt>,
like:
<pre class="config">
Contents:
</pre>
Versions before 3.0.0 need a ratio number there, like:
<pre class="config">
Contents: 1
</pre>
The number is the inverse ratio of not yet looked at and cached files to process in
every run. The larger the more packages are missing. 1 means to list everything.
<br>
The arguments of the Contents field and other fields control
which Architectures to generate Contents files for and which
Components to include in those. For example
<pre class="config">
Contents: udebs nodebs . .gz .bz2
ContentsArchitectures: ia64
ContentsComponents:
ContentsUComponents: main
</pre>
means to not skip any packages, generate Contents for <tt class="suffix">.udeb</tt>
files, not generating Contents for <tt class="suffix">.deb</tt>s. Also it is only
generated for the <tt>ia64</tt> architecture and only packages in component
<tt>main</tt> are included.
<h4><a name="exporthook">Additional index files (like .diff)</a></h4>
Index files reprepro cannot generate itself, can be generated by telling
it to call a script.
<h5>using rredtool to generate pdiff files</h5>
Starting with version 4.1.0, the <tt>rredtool</tt> coming with reprepro
can be used as hook to create and update <tt>Packages.diff/Index</tt> files.
<br>
Unlike dak (which created the official Debian repositories) or the tiffany.py
script (see below) derived from dak, an user will only need to download
one of those patches, as new changes are merged into the old files.
<br>
To use it, make sure you have
<a href="http://packages.debian.org/diff">diff</a> and
<a href="http://packages.debian.org/gzip">gzip</a>
installed.
Then add something like the following to the headers of the distributions
that should use this in <tt class="file">conf/distributions</tt>:
<pre class="config">
 DscIndices: Sources Release . .gz /usr/bin/rredtool
 DebIndices: Packages Release . .gz /usr/bin/rredtool
</pre>
<h5>the tiffany example hook script (generates pdiff files)</h5>
This example generates <tt class="file">Packages.diff</tt> and/or
<tt class="file">Sources.diff</tt> directories containing a set of
ed-style patches, so that people do not redownload the whole index
for just some small changes.
<br>
To use it, copy <tt class="file">tiffany.example</tt> from the examples directory
into your <tt class="dir">conf</tt> directory.
(or any other directory, then you will need to give an absolute path later).
Unpack, if needed. Rename it to tiffany.py and make it executeable.
Make sure you have
<a href="http://packages.debian.org/python-apt">python-apt</a>,
<a href="http://packages.debian.org/diff">diff</a> and
<a href="http://packages.debian.org/gzip">gzip</a>
installed.
Then add something like the following to the headers of the distributions
that should use this in <tt class="file">conf/distributions</tt>:
<pre class="config">
 DscIndices: Sources Release . .gz tiffany.py
 DebIndices: Packages Release . .gz tiffany.py
</pre>
More information can be found in the file itself. You should read it.
<h5>the bzip2 example hook script</h5>
This is an very simple example.
Simple and mostly useless,
as reprepro has built in <tt>.bz2</tt> generation support,
unless you compiled it your own with <tt>--without-libbz2</tt> or
with no <tt>libbz2-dev</tt> installed.
<br>
To use it, copy <tt class="file">bzip.example</tt> from the examples directory
into your <tt class="dir">conf</tt> directory.
(or any other directory, then you will need to give an absolute path later).
Unpack, if needed. Rename it to bzip2.sh and make it executeable.
Then add something like the following to the headers of the distributions
that should use this in <tt class="file">conf/distributions</tt>:
<pre class="config">
 DscIndices: Sources Release . .gz bzip2.sh
 DebIndices: Packages Release . .gz bzip2.sh
 UDebIndices: Packages . .gz bzip2.sh
</pre>
The script will compress the index file using the
<a href="http://packages.debian.org/bzip2">bzip2</a> program and tell
reprepro which files to include in the Release file of the distribution.
<h5>internals</h5>
TO BE CONTINUED
<h4>...</h4>
TO BE CONTINUED
<h2><a name="localpackages">Local packages</a></h2>
There are two ways to get packages not yet in any repository into yours.
<dl><dt>includedsc, includedeb, include</dt><dd>
These are for including packages at the command line.
Many options are available to control what actually happens.
You can easily force components, section and priority and/or choose to
include only some files or only in specific architectures.
(Can be quite usefull for architecture all packages depending on some
packages you will some time before building for some of your architectures).
Files can be moved instead of copied and most sanity checks overwritten.
They are also optimized towards being fast and simply try things instead of
checking a long time if they would succeed.
</dd><dt>processincoming</dt><dd>
This command checks for changes files in an incoming directory.
Being optimized for automatic processing (i.e. trying to checking
everything before actually doing anything), it can be slower
(as every file is copied at least once to sure the owner is correct,
with multiple partitions another copy can follow).
Component, section and priority can only be changed via the distribution's
override files. Every inclusion needs a <tt class="suffix">.changes</tt> file.
<br>
This method is also relatively new (only available since 2.0.0), thus
optimisation for automatic procession will happen even more.
</dd></dl>
<h3><a name="include">Including via command line</a></h3>
There are three commands to directly include packages into your repository:
<tt class="command">includedeb</tt>, <tt class="command">includedsc</tt>
and <tt class="command">includechanges</tt>.
Each needs to codename of the distribution you want to put your package into
as first argument and a file of the appropiate type
(<tt class="suffix">.deb</tt>, <tt class="suffix">.dsc</tt> or
 <tt class="suffix">.changes</tt>, respectively) as second argument.
<br>
If no component is specified via <tt class="option">--component</tt>
(or short <tt class="option">-C</tt>), it will be guessed looking at its
section and the components of that distribution.
<br>
If there are no <tt class="option">--section</tt>
(or short <tt class="option">-S</tt>) option, and it is not specified
by the (binary or source, depending on the type) override file of the
distribution, the value from the <tt class="suffix">.changes</tt>-file
is used (if the command is <tt class="command">includechanges</tt>)
or it is extracted out of the file (if it is a
<tt class="suffix">.deb</tt>-file, future versions might also try to
extract it from a <tt class="suffix">.dsc</tt>'s diff or tarball).
<br>
Same with the priority and the <tt class="option">--priority</tt>
(or short <tt class="option">-P</tt>) option.
<br>
With the <tt class="option">--architecture</tt> (or short <tt class="option">-A</tt>)
option, the scope of the command is limited to that architecture.
<tt class="command">includdeb</tt> will add a Architecture <tt>all</tt>
packages only to that architecture (and complain about Debian packages for
other architectures).
<tt class="command">include</tt> will do the same and ignore packages for
other architectures (source packages will only be included if the value
for <tt class="option">--architecture</tt> is <tt>source</tt>).
<br>
To limit the scope to a specify type of package, use the
<tt class="option">--packagetype</tt> or short <tt class="option">-T</tt>
option. Possible values are <tt>deb</tt>, <tt>udeb</tt> and <tt>dsc</tt>.
<br>
When using the <tt class="option">--delete</tt> option, files will
be moved or deleted after copying them.
Repeating the <tt class="option">--delete</tt> option will also delete
unused files.
<br>
TO BE CONTINUED.
<h3><a name="incoming">Processing an incoming queue</a></h3>
Using the <tt class="command">processincoming</tt> command reprepro
can automatically process incoming queues.
While this is still improveable (reprepro still misses ways to send
mails and especially an easy way to send rejection mails to the
uploader directly), it makes it easy to have an directory where you
place your packages and reprepro will automatically include them.
<br>
To get this working you need three things:
<ul>
<li><a href="#processincoming-incoming-config">
a file <tt class="file">conf/incoming</tt> describing your incoming directories,
</a></li>
<li><a href="#processincoming-dist-config">
a <tt class="file">conf/distribution</tt> file describing your distributions
(as always with reprepro)
and
</a></li>
<li><a href="#processincoming-calling">
a way to get reprepro called to process it.
</a></li>
</ul>
<a name="processincoming-incoming-config">
<h4>The file <tt class="file">conf/incoming</tt></h4></a>
describes the different incoming queues.
As usual the different chunks are separated by empty lines.
Each chunk can have the following fields:
<dl><dt>Name</dt><dd>This
is the name of the incoming queue, that <tt class="command">processincoming</tt>
wants as argument.</dd>
<dt>IncomingDir</dt><dd>The actual directory to look for
<tt class="suffix">.changes</tt> files.</dd>
<dt>TempDir</dt><dd>To ensure integrity of the processed files and their
permissions,
every file is first copied from the incoming directory to this directory.
Only the user reprepro runs as needs write permissions here.
It speeds things up if this directory is in the same partition as the pool.
<dt>Allow</dt><dd>
This field lists the distributions this incoming queue might inject packages
into.
Each item can be a pair of a name of a distribution to accept and a distribution
to put it into.
Each upload has each item in its <tt class="field">Distribution:</tt> field
compared first to last to each of this items and is put in the first distribution
accepting it. For example
<pre class="line">
Allow: stable>etch stable>etch-proposed-updates mystuff unstable>sid
</pre>
will put a <tt class="suffix">.changes</tt> file with
<tt class="field">Distribution: stable</tt> into etch.
If that is not possible (e.g. because etch has a
<tt class="field">UploadersList</tt> option not allowing this) it will
be put into etch-proposed-updates.
And a <tt class="suffix">.changes</tt> file with
<tt class="field">Distribution: unstable</tt> will be put into sid, while
with <tt class="field">Distribution: mystuff</tt> will end up in mystuff.
<br>
If there is a <tt class="field">Default</tt> field, the <tt class="field">Allow</tt>
field is optional.</dd>
<dt>Default</dt><dd>
Every upload not catched by an item of the <tt class="field">Allow</tt>
field is put into the distribution specified by this.
<br>
If there is a <tt class="field">Allow</tt> field, the <tt class="field">Default</tt>
field is optional.</dd>
<dt>Multiple</dt><dd>
This field only makes a difference if a <tt class="suffix">.changes</tt> file
has multiple distributions listed in its <tt class="field">Distribution:</tt>
field.
Without this field each of those distributions is tried according to the
above rules until the package is added to one (or none accepts it).
With this field it is tried for each distribution, so a package can be upload
to multiple distributions at the same time.
</dd>
<dt>Permit</dt><dd>
A list of options to allow things otherwise causing errors.
(see the manpage for possible values).
<br>This field os optional.</dd>
<dt>Cleanup</dt><dd>
Determines when and what files to delete from the incoming queue.
By default only sucessfully processed <tt class="suffix">.changes</tt> files
and the files references by those are deleted.
For a list of possible options take a look into the man page.
<br>This field os optional.</dd>
</dl>
<a name="processincoming-dist-config">
<h4><tt class="file">conf/distribution</tt> for <tt class="command">processincoming</tt></h4></a>
There are no special requirements on the <tt class="file">conf/distribution</tt>
file by processincoming. So even a simple
<pre class="file">
Codename: mystuff
Architectures: i386 source
Components: main non-free contrib bad
</pre>
will work.
<br>
The <tt class="field">Uploaders</tt> field can list a file limiting
uploads to this distribution to specific keys and
<tt class="field">AlsoAcceptFor</tt> is used to resolve unknown names
in <tt class="file">conf/incoming</tt>'s <tt class="field">Allow</tt>
and <tt class="field">Default</tt> fields.
<a name="processincoming-calling">
<h4>Getting <tt class="command">processincoming</tt> called.</h4></a>
While you can just call <tt class="command">reprepro processincoming</tt> manually,
having an incoming queue needing manual intervention takes all the fun out of
having an incoming queue, so usually so automatic way is choosen:
<ul>
<li>Dupload and dput have ways to call an hook after an package was uploaded.
This can be an ssh to the host calling reprepro.
The disavantage is having to configure this in every
<tt class="file">.dupload.conf</tt> on every host you want to upload and give
everyone access to ssh and permissions on the archive who should upload.
The advantage is you can configure reprepro to have interactive scripts or
ask for passphrases.
</li>
<li>Install a cron-job calling reprepro every 5 minutes. Cron is usually
available everywhere and getting the output sent by mail to you or a mailing
list is easy.
The annoying part is having to wait almost 5 minutes for the processing.
</li>
<li>Use something like <a href="http://packages.debian.org/inoticoming"><tt class="external">inoticoming</tt></a>.
Linux has a syscall called inotify, allowing a program to be run whenever
something happens to a file.
One program making use of this is inoticoming. It watches a directory using
this facility and whenever a <tt class="suffix">.changes</tt> file is completed
it can call reprepro for you.
(As this happens directly, make sure you always upload the <tt class="suffix">.changes</tt>
file last, dupload and dput always ensure this).
This can be combined with Debian's cron-extension to have a program started at
boot time with the <tt>@reboot</tt> directive.
For example with a crontab like:
<pre class="file">
MAILTO=myaddress@somewhere.tld

@reboot inoticoming --logfile /my/basedir/logs/i.log /my/basedir/incoming/ --stderr-to-log --stdout-to-log --suffix '.changes' --chdir /my/basedir reprepro -b /my/basedir --waitforlock 100 processincoming local {} \;
</pre>
</li>
</ul>
<h2><a name="mirroring">Mirroring / Updating</a></h2>
Reprepro can fetch packages from other repositories.
For this it uses apt's methods from <tt class="dir">/usr/lib/apt/methods/</tt>
so everything (http, ftp, ...) that works with apt should also work with reprepro.
Note that this works on the level of packages, even though you can tell reprepro
to create a distribution having always the same packages as some remote repository,
the repository as a whole may not look exactly the same but only have the same set
of packages in the same versions.
<br>
You can also only mirror a specific subset of packages, merge multiple repositories
into one distribution, or even have distributions mixing remote and local packages.
<br>
Each distribution to receive packages from other repositories needs an
<tt class="field">Update:</tt> field listing the update rules applied to it.
Those update rules are listed in <tt class="file">conf/updates</tt>.
There is also the magic <tt>-</tt> update rule, which tells reprepro to delete
all packages not readded by later rules.
<br>
To make reprepro to update all distributions call <tt>reprepro update</tt>
without further arguments, or give the distributions to update as additional
arguments.
<br>
Let's start with some examples:
<h3><a name="update-examples">Updating examples</a></h3>
Let's assume you have the following <tt class="file">conf/distributions</tt>
<pre class="file">
Codename: etch
Architectures: i386 source
Components: main contrib
Update: local - debian security

Codename: mystuff
Architectures: abacus source
Components: main bad
Update: debiantomystuff
</pre>
and the following <tt class="file">conf/updates</tt>
<pre class="file">
Name: local
Method: http://ftp.myorg.tld/debian

Name: debian
Method: http://ftp.de.debian.org/debian
VerifyRelease: A70DAF536070D3A1
Config: Acquire::Http::Proxy=http://proxy.yours.org:8080

Name: security
Suite: */updates
Method: http://security.eu.debian.org/
Fallback: http://security.debian.org/
VerifyRelease: A70DAF536070D3A1
Config: Acquire::Http::Proxy=http://proxy.yours.org:8080

Name: debiantomystuff
Suite: sid
Method: http://ftp.de.debian.org/debian
Architectures: i386&gt;abacus source
Components: main non-free&gt;bad contrib&gt;bad
FilterFormula: Architecture (== all)| !Architecture
FilterList: deinstall list
</pre>
and a file <tt class="file">conf/list</tt> with some
output as <tt>dpkg --get-selections</tt> is printing.
<br>
If you then run
<tt class="command">reprepro update etch</tt> or
<tt class="command">reprepro checkupdate etch</tt>,
reprepro looks at etch's <tt class="field">Update:</tt> line
and finds four rules. The first is the <tt>local</tt> rule,
which only has a method, so that means it will download the
<tt class="file">Release</tt> file from
<tt>http://ftp.myorg.tld/debian/dists/etch/Release</tt> and
(unless it already has downloaded them before or that
repository does not have all of them) downloads the
<tt>binary-i386/Packages.gz</tt>
and <tt>source/Sources.gz</tt> files for main and contrib.
The same is done for the <tt>debian</tt> and <tt>security</tt>
rules.
As they have a <tt class="field">VerifyRelease</tt> field,
Release.gpg is also downloaded and checked to be signed with the
given key
(which you should have imported to you <tt class="external">gpg</tt>
keyring before).
As security has a <tt class="field">Suite:</tt> field, not the codename,
but the content of this field (with an possible<tt>*</tt> replaced by the codename),
is used as distribution to get.
<br>
Then it will parse for each part of the distribution, parse the files it
get from left to right.
For each package it starts with the version currently in the distribution,
if there is a newer on in <tt>local</tt> it will mark this.
Then there is the delete rule <tt>-</tt>, which will mark it to be deleted
(but remembers what was there, so if later the version in the distribution
or the version in <tt>local</tt> are newest, it will get them from here avoiding
slow downloads from far away). Then it will look into <tt>debian</tt> and then
in <tt>security</tt>, if they have a newer version (or the same version, clearing
the deletion mark).
<br>
If you issued <tt class="command">checkupdate</tt> reprepro will print what it would
do now, otherwise it tries to download all the needed files and when it got all,
change the packages in the distribution to the new ones, export the index files
for this distribution and finaly delete old files no longer needed.
<br>
TO BE CONTINUED.
<h2><a name="propagation">Propagation of packages</a></h2>
You can copy packages between distributions using the
<tt class="command">pull</tt> and <tt class="command">copy</tt> commands.
<br>
With the <tt class="command">copy</tt> command you can copy packages
by name from one distribution to the other within the same repository.
<br>
With the <tt class="command">pull</tt> command you can pull all packages
(or a subset defined by some list, or exceptions by some list, or by some
formula, or ...) from one distribution to another within the same formula.
<br>
Note that both assume the filenames of the corresponding packages in the
pool will not differ, so you cannot move packages from one component to another.
<br>
Let's just look at a little example, more information can be found in the man page.
<br>
Assume you upload all new packages to a distribution and you want another
so you can keep using an old version until you know the newer works, too.
One way would be to use something like the following
<tt class="file">conf/distributions</tt>:
<pre class="file">
Codename: development
Suite: unstable
Components: main extra
Architectures: i386 source

Codename: bla
Suite: testing
Components: main extra
Architectures: i386 source
Pull: from_development
</pre>
and <tt class="file">conf/pulls</tt>:
<pre class="file">
Name: from_development
From: development
</pre>
i.e. you have two distributions, bla and development.
Now you can just upload stuff to development (or it's alias unstable).
And when you want a single package to go to testing, you can use the copy
command:
<pre class="shell">
reprepro copy bla development name1 name2 name3
</pre>
If you do not want to copy all pakages of a given name, but only some
of them, you can use <tt>-A</tt>, <tt>-T</tt> and <tt>-C</tt>:
<pre class="shell">
reprepro -T deb -A i386 copy bla development name1
</pre>
will copy <tt class="suffix">.deb</tt> packages called name1 from the i386
parts of the distribution.
<br>
TO BE CONTINUED
<h2><a name="snapshosts">Snapshots</a></h2>
There is a gensnapshot command.<br>
TO BE DOCUMENTED
<h2><a name="tracking">Source package tracking</a></h2>
TO BE DOCUMENTED
<h2><a name="hooks">Extending reprepro / Hooks and more</a></h2>
When reprepro misses some functionality,
it often can be be added by some kind of hook.
<br>
Currently you can execute your own scripts at the following occasions:
<ul>
<li><a href="#addhook">after adding or removing packages</a></li>
<li><a href="#byhandhook">to process byhand files</a></li>
<li><a href="#exporthook">when creating index files (Packages.gz, Sources.gz)</a></li>
<li><a href="#signhook">when signing releases</a></li>
<li><a href="#outhook">after changing the visible files of the repository managed</a></li>
<li><a href="#endhook">when reprepro finished</a></li>
</ul>
<h3><a name="addhook">Scripts to be run when adding or removing packages</a></h3>
Whenever a package is added or removed,
you can tell reprepro to log that to some file and/or call a script using the
<tt>Log:</tt> directive in <tt class="file">conf/distributions</tt>.
<br>
This script can send out mails and do other logging stuff,
but despite the name, it is not restricted to logging.
<br>
<h4>Automatically extracting changelog and copyright information</h4>
reprepro ships with an example script to extract <tt class="file">debian/changelog</tt>
and <tt class="file">debian/copyright</tt>
files from source packages into a hierachy loosely resembling the way changelogs
are made available at
<a href="http://packages.debian.org/changelogs/">http://packages.debian.org/changelogs/</a>.
<br>
All you have to do is to copy (or unpack if compressed) the file
<tt class="file">changelogs.example</tt> from the examples directory
in the reprepro source or
<a href="file:///usr/share/doc/reprepro/examples/">/usr/share/doc/reprepro/examples/</a>
of your installed reprepro package into your <tt class="directory">conf/</tt> directory
(or somewhere else, then you will need an absolute path later), perhaps
change some directories specified in it
and add something like the following lines
to all distributions in <tt class="file">conf/distributions</tt> that should use
this feature:
<pre class="config">
Log:
 --type=dsc changelogs.example
</pre>
If you still want to log to some file, just keep the filename there:
<pre class="config">
Log: mylogfilename
 --type=dsc changelogs.example
</pre>
Then cause those files to be generated for all existing files via
<pre class="command">
reprepro rerunnotifiers
</pre>
and all future source packages added or removed will get this list automatically
updated.
<h4>Writing your own Log: scripts</h4>
You can list an arbitrary amount of scripts, to be called at specified times
(which can overlap or even be the same):
<pre class="config">
Log: logfilename
 --type=dsc script-to-run-on-source-package-changes
 script-to-run-on-package-changes
 another-script-to-run-on-package-changes
 --type=dsc --component=main script-to-run-on-main-source-packages
 --architecture=i386 --type=udeb script-to-run-on-i386-udebs
 --changes script-to-run-on-include-or-processincoming
</pre>
There are two kind of scripts:
The first one is called when a package was added or removed.
Using the <tt class="option">--archtecture=</tt>,
<tt class="option">--component=</tt> and
<tt class="option">--type=</tt> options you can limit it to specific parts
of the distribution.
The second kind is marked with <tt class="option">--changes</tt> and is
called when a <tt class="suffix">.changes</tt>-file was added with
<tt class="command">include</tt> or <tt class="command">processincoming</tt>.
Both are called asynchonous in the background <emph>after</emph> everything was done,
but before no longer referenced files are deleted (so the files of the
replaced or deleted package are still around).
<h5>Calling conventions for package addition/removal scripts</h5>
This type of script is called with a variable number of arguments.
The first argument is the action. This is either
<tt>add</tt>, <tt>remove</tt> or <tt>replace</tt>.
The next four arguments are the codename of the affected distribution
and the packagetype, component and architecture in that distribution
affected.
The sixth argument is the package's name.
After that is the version of the added package (<tt>add</tt> and <tt>replace</tt>)
and the version of the removed package (<tt>remove</tt> and <tt>replace</tt>).
Finally the filekeys of the new (<tt>add</tt> and <tt>replace</tt>) and/or
removed (<tt>remove</tt> and <tt>replace</tt>) package are listed
starting with the marker &quot;<tt>--</tt>&quot; followed by each filekey
(the name of the file in the <tt class="dir">pool/</tt>
relative to <tt class="env">REPREPRO_OUT_DIR</tt>)
as its own argument.
<br>
The environment variable <tt class="env">REPREPRO_CAUSING_COMMAND</tt>
contains the command of the action causing this change.
The environment variable
<tt class="env">REPREPRO_CAUSING_FILE</tt> contains the name of the file
given at the command line causing this package to be changed,
if there is one.
(i.e. with <tt class="command">includedeb</tt>,
<tt class="command">includedsc</tt> and <tt class="command">include</tt>).
The environment variables
<tt class="env">REPREPRO_CAUSING_RULE</tt> and
<tt class="env">REPREPRO_FROM</tt> are
the name of the update or pull rule pulling in a package
and the name of the distribution a package is coming from.
What this name is depends on the command and for most commands
it is simply not set at all.
And of course all the <tt class="env">REPREPRO_*_DIR</tt> variables are set.
<h5>Calling conventions for <tt class="suffix">.changes</tt> scripts</h5>
This type of script is called with 5 or 6 arguments.
The first is always &quot;<tt>accepted</tt>&quot;, to make it easier to
check it is configued the right way.
The second argument is the codename of the distribution the
<tt class="suffix">.changes</tt>-file was added to.
The third argument is the source name, the forth the version.
The fifth name is the <tt class="suffix">.changes</tt> itself
(in case of <tt class="command">processingcoming</tt> the secure copy in the
temporary dir).
There is a sixth argument if the <tt class="suffix">.changes</tt>-file was
added to the <tt class="dir">pool/</tt>:
The filekey of the added .changes file
(i.e. the filename relative to <tt class="env">REPREPRO_OUT_DIR</tt>).
<br>
The environment variable <tt class="env">REPREPRO_CAUSING_COMMAND</tt>
contains the command of the action causing this change.
The environment variable
<tt class="env">REPREPRO_CAUSING_FILE</tt> contains the name of the file
given at the command line, if there is one
(e.g. with <tt class="command">include</tt>).
And of course all the <tt class="env">REPREPRO_*_DIR</tt> variables are set.
<h3><a name="byhandhook">Scripts to be run to process byhand files</a></h3>
<tt class="suffix">.changes</tt> files can (beside the usual packages files
to be included in the repository) contain additional files to be processed
specially.
Those are marked by the special section <tt class="constant">byhand</tt> (in Debian)
or <tt class="constant">raw-</tt>something (in Ubuntu).
Besides storing them just in the pool besides the packages using the
<tt class="constant">includebyhand</tt> value in the <tt class="field">Tracking</tt>
settings you can also let reprepro process a hook to process them when encountering
them in the <tt class="action">processincomming</tt> action
(Typical usages are uploading documentation files this way that are unpacked next
to the repository, or installer images or stuff like that).

To use them add to the distribution's defining stanca in <tt class="filename">conf/distributions</tt> a field like:
<pre class="config">
ByhandHooks:
 byhand * manifesto.txt handle-byhand.sh
</pre>
This will call the hook script <tt class="constant">handle-byhand.sh</tt> for every byhand file with section <tt class="constant">byhand</tt>, any priority and filename <tt class="constant">manifesto.txt</tt>. (The first three fields allow glob characters for matching).

The script will then be alled with 5 arguments: the codename of the distribution, the section, the priority, the filename as found in the changes file and the filename of where the script can find the actual file.
<h3>Scripts to be run when creating index files (Packages.gz, Sources.gz)</h3>
this hook is described in the section <a href="#exporthook">&quot;Additional index files&quot;</a>.
<h3><a name="signhook">Scripts to be run when signing releases</a></h3>
Instead of creating <tt class="filename">InRelease</tt> and
<tt class="filename">Release.gpg</tt> files using libgpgme,
the <tt class="option">SignWith</tt> option can also contain
a exclamation mark followed by a space and the name of a hook script to call.

The script gets three arguments: The filename to sign, the filename of the InRelease
file to create and the filename of the Release.gpg to create (a Release.gpg does not
need to be created. reprepro will assume you do not care about that legacy file if it
<h3><a name="outhook">Scripts to be run after changing the visible files of the repository managed</a></h3>
When using the <tt class="option">--outhook</tt> command line option (or the corresponding
<tt class="constant">outhook</tt> in the <tt class="filename">options</tt> file),
reprepro will create a <tt class="suffix">.outlog</tt> file in the log directory describing
any changes done to the out dir and calls the hook script given as argument with this
file as argument.

The <tt class="suffix">.outlog</tt> file consists of lines each starting with a keyword
and then some arguments seperated by tab characters.

The possible keywords are:
<ul>
<li><tt class="constant">POOLNEW</tt>:
One argument is the filekey of a file newly added to the pool.
<li><tt class="constant">POOLDELETE</tt>:
One argument is the filekey of a file removed from the pool.
<li><tt class="constant">START-DISTRIBUTION</tt>, <tt class="constant">END-DISTRIBUTION</tt>:
two or three arguments: the codename, the directory,
and the suite (if set).
<li><tt class="constant">START-SNAPSHOT</tt>, <tt class="constant">END-SNAPSHOT</tt>:
three arguments: the codename, the directory, and the name of the snapshot generated.
<li><tt class="constant">DISTFILE</tt>:
three arguments: the directory of the distribution (relative to out dir), the name relative to that directory, and the filename generated by reprepro.
<li><tt class="constant">DISTSYMLINK</tt>:
three arguments: the directory of the distribution (relative to out dir), the name relative to that directory, and the symlink target (relative to that directory).
<li><tt class="constant">DISTDELETE</tt>:
two arguments: the directory of the distribution (relative to out dir), the name relative to that directory of a file no longer there.
<li><tt class="constant">DISTKEEP</tt> (not yet generated):
two arguments: the directory of the distribution (relative to out dir), the name relative to that directory.
</ul>

All <tt class="constant">POOLNEW</tt> come before any distribution changes referencing them
and all <tt class="constant">POOLDELETE</tt> will be afterwards.
Each line beloging to a distribution is guaranteed to be between the corresponding
<tt class="constant">START-DISTRIBUTION</tt> and
<tt class="constant">END-DISTRIBUTION</tt> or between a
<tt class="constant">START-SNAPSHOT</tt> and
<tt class="constant">END-SNAPSHOT</tt> or between a
with the same directory
(i.e. there is some redundancy so you can choose to parse the information where it is more conventient for you).

The lines starting with <tt class="constant">DIST</tt> describe new or modified files in the distribution description exported by reprepro. No hint is given if that file was previously non-existant, a proper file or a symlink (i.e. if you copy stuff, do not make any assumptions about that).
Future versions of reprepro might create <tt class="constant">DISTKEEP</tt> lines to denote files that have not changed (i.e. just ignore those lines to be future-proof).

The directories for the distribution entries are what apt expects them (i.e. always starting with <tt class="constant">dists/</tt>, while the third argument to <tt class="constant">DISTFILE</tt> is the name reprepro generated (i.e. starts with the distdir value, which can be configured to not end with <tt class="constant">dists/</tt>).

<h3><a name="endhook">when reprepro finished</a></h3>
With the <tt class="option">--endhook</tt> command line option (or the corresponding
<tt class="constant">endhook</tt> in the <tt class="filename">options</tt> file) you
can specify a hook to be executed after reprepro finished but before reprepro returns
the to calling process.
The hook gets all the command line arguments after the options (i.e. starting with
the name of the action) and the exit code reprepro would have produces.
For an example see the man page.
<h2><a name="maintenance">Maintenance</a></h2>
This section lists some commands you can use to check and improve the health
of you repository.
<br>
Normally nothing of this should be needed, but taking a look from time to time
cannot harm.
<pre class="command">
reprepro -b $YOURBASEDIR dumpunreferenced
</pre>
This lists all files reprepro knows about that are not marked as needed by anything.
Unless you called reprepro with the <tt class="option">--keepunreferenced</tt>
option, those should never occour. Though if reprepro is confused or interupted it
may sometimes prefer keeping files around instead of deleting them.
<pre class="command">
reprepro -b $YOURBASEDIR deleteunreferenced
</pre>
This is like the command before, only that such files are directly forgotten and
deleted.
<pre class="command">
reprepro -b $YOURBASEDIR check
</pre>
Look if all needed files are in fact marked needed and known.
<pre class="command">
reprepro -b $YOURBASEDIR checkpool
</pre>
Make sure all known files are still there and still have the same checksum.
<pre class="command">
reprepro -b $YOURBASEDIR checkpool fast
</pre>
As the command above, but do not compute checksums.
<pre class="command">
reprepro -b $YOURBASEDIR tidytracks
</pre>
If you use source package tracking, check for files kept because of this
that should no longer by the current rules.
<br>
If you fear your tracking data could have became outdated,
you can also try the retrack command:
<pre class="command">
reprepro -b $YOURBASEDIR retrack
</pre>
That refreshes the tracking information about packages used and then
runs a tidytracks. (Beware: don't do this with reprepro versions before 3.0.0).
<h2><a name="internals">Internals</a></h2>
reprepro stores the data it collects in Berkeley DB file (<tt class="suffix">.db</tt>)
in a directory called <tt class="dir">db/</tt> or whatever you specified via command
line. With a few exceptions, those files are NO CACHES, but the actual data.
While some of those data can be regained when you lose those files, they are better
not deleted.
<h3>packages.db</h3>
This file contains the actual package information.
<br>
It contains a database for every (codename,component,architecture,packagetype) quadruple
available.
<br>
Each is indexed by package name and essentially contains the information written do
the Packages and Sources files.
<br>
Note that if you change your <tt class="file">conf/distributions</tt> to no longer
list some codenames, architectures or components,
that will not remove the associated databases in this file.
That needs an explicit call to <tt class="command">clearvanished</tt>.
<h3>references.db</h3>
This file contains a single database that lists for every file why this file
is still needed.
This is either an identifier for a package database, an tracked source package,
or a snapshot.
<br>
Some low level commands to access this are (take a look at the manpage for how to use them):
<dl class="commands">
<dt class="command">rereference</dt><dd>recreate references (i.e. forget old and create newly)</dd>
<dt class="command">dumpreferences</dt><dd>print a list of all references</dd>
<dt class="command">_removereferences</dt><dd>remove everything referenced by a given identifier</dd>
<dt class="command">_addreference</dt><dd>manually add a reference</dd>
</dl>
<h3>files.db / checksums.db</h3>
These files contains what reprepro knows about your <tt class="dir">pool/</tt> directory,
i.e. what files it things are there with what sizes and checksums.
The file <tt class="filename">files.db</tt> is used by reprepro before version 3.3
and kept for backwards compatibility.
If your repository was only used with newer versions you can safely delete it.
Otherwise you should run <tt class="command">collectnewchecksums</tt> before deleting
it.
The file <tt class="filename">checksums.db</tt> is the new file used since
version 3.3.
It can store more checksums types (<tt class="filename">files.db</tt> only contained
md5sums, <tt class="filename">checksums.db</tt> can store arbitrary checksums and
reprepro can even cope with it containing checksum types it does not yet know of)
but for compatibility with pre-3.3 versions is not the canonical source of information
as long as a <tt class="filename">files.db</tt> file exists).
<br>
If you manually put files in the pool or remove them, you should tell reprepro about that.
(it sometimes looks for files there without being told, but it never forgets files
except when it would have deleted them anyway).
Some low level commands (take a look at the man page for how to use them):
<dl class="commands">
<dt class="command">collectnewchecksums</dt><dd>Make sure every file is listed in <tt class="filename">checksums.db</tt> and with all checksum types your reprepro supports.</dd>
<dt class="command">checkpool fast</dt><dd>Make sure all files are still there.</dd>
<dt class="command">checkpool</dt><dd>Make sure all files are still there and correct.</dd>
<dt class="command">dumpunreferenced</dt><dd>Show all known files without reference.</dd>
<dt class="command">deleteunreferenced</dt><dd>Delete all known files without reference.</dd>
<dt class="command">_listmd5sums</dt><dd>Dump this database (old style)</dd>
<dt class="command">_listchecksums</dt><dd>Dump this database (new style)</dd>
<dt class="command">_detect</dt><dd>Add files to the database</dd>
<dt class="command">_forget</dt><dd>Forget that some file is there</dd>
<dt class="command">_addmd5sums</dt><dd>Create the database from dumped data</dd>
<dt class="command">_addchecksums</dt><dd>dito</dd>
<h3>release.cache.db</h3>
In this file reprepro remembers what it already wrote to the <tt class="dir">dists</tt>
directory,
so that it can write their checksums (including the checksums of the uncompressed variant,
even if that was never written to disk)
in a newly to create <tt class="file">Release</tt>
file without having to trust those files or having to unpack them.
<h3>contents.cache.db</h3>
This file contains all the lists of files of binary package files where reprepro
already needed them. (which can only happen if you requested Contents files to be
generated).
<h3>tracking.db</h3>
This file contains the information of the <a href="#tracking">source package tracking</a>.
<h2><a name="recovery">Disaster recovery</a></h2>
TO BE DOCUMENTED (see the
<a href="http://git.debian.org/?p=mirrorer/reprepro.git;a=blob_plain;f=docs/recovery;hb=HEAD">recovery</a>
file until then)
<h2><a name="paranoia">Paranoia</a></h2>
As all software, reprepro might have bugs.
And it uses libraries not written by myself,
which I'm thus even more sure that they will have bugs.
Some of those bugs might be security relevant.
This section contains some tips, to reduce the impact of those.
<ul>
<li>Never run reprepro as root.<br>
All reprepro needs to work are permissions to files,
there is no excuse for running it as root.
</li>
<li>Don't publish your db/ directory.<br>
The contents of the db directory are not needed by everyone else.
Having them available to everyone may make it easier for them to
exploit some hypothetical problem in libdb and makes it easier to
know in advance how exactly reprepro will act in a given circumstances,
thus easier to exploit some hypothetical problem.
</li>
<li>Don't accept untrusted data without need.<br>
If an attacker cannot do anything, they cannot do anything harmful, either.
So if there is no need, don't offer an anonymous incoming queue.
<tt class="program">dput</tt> supports uploading via scp, so just having
an only group-writable incoming directory, or even better multiple incoming
directories can be a better alternative.
</li>
</ul>
External stuff being used and attack vectors opened by it:
<dl>
<dt>libgpgme/gpg</dt><dd>
Almost anything is run through <tt>libgpgme</tt> and thus <tt>gpg</tt>.
It will be used to check the <tt class="filename">Release.gpg</tt> file,
or to read <tt class="suffix">.dsc</tt> and <tt class="suffix">.changes</tt>
files (even when there is no key to look for specified,
as that is the best way to get the data from the signed block).
Avoiding this by just accepting stuff without looking for signatures on
untrusted data is not really an option, so I know nothing to prefent this
type of problems.
</dd>
<dt>libarchive</dt><dd>
The <tt class="suffix">.tar</tt> files within <tt class="suffix">.deb</tt>
files are normaly (unless that library was
not available while compiling) read using libarchive.
This happens when a <tt class="suffix">.deb</tt> file is to be added
(though only after deciding if it should be added, so if it does not have
the correct checksum or the .changes did not have the signatures you specified,
it is not) or when the file list is to be extracted
(when creating <tt class="filename">Contents</tt> files).
Note that they are not processed when only mirroring them (of course unless
<tt class="filename">Contents</tt> files are generated), as then only the
information from the Packages file is copied.
</dd>
<dt>dpkg-deb/tar</dt><dd>
If reprepro was compiled without libarchive,
<tt class="program">dpkg-deb</tt> is used instead, which most likely will
call <tt class="program">tar</tt>. Otherwise just the same like the last
item.
</dd>
<dt>zlib</dt><dd>
When mirroring packages, the downloaded
<tt class="filename">Packages.gz</tt> and <tt class="filename">Sources.gz</tt> files
are read using zlib. Also the generated <tt class="suffix">.gz</tt> files
are generated using it. There is no option but hoping there is no security
relevant problem in that library.
</dd>
<dt>libbz2</dt><dd>
Only used to generate <tt class="suffix">.bz2</tt> files.
If you fear simple blockwise writing using that library has a security problem
that can be exploited by data enough harmless looking to be written to the
generated index files, you can always decide to no tell reprepro to generate
<tt class="suffix">.bz2</tt> files.
</dd>
</dl>
<h2><a name="counterindications">What reprepro cannot do</a></h2>
There are some things reprepro does not do:
<dl><dt>Verbatim mirroring</dt><dd>
Reprepro aims to put all files into a coherent <tt>pool/</tt> hierarchy.
Thus it cannot guarantee that files will have the same relatives path as in the
original repository (especially if those have no pool).
It also creates the index files from its own indices.
While this leads to a tidy repository and possible savings of disk-space, the
signatures of the repositories you mirror cannot be used to authenticate the mirror,
but you will have to sign (or tell reprepro to sign for you) the result.
While this is perfect when you only mirror some parts or specific packages or
also have local packages that need local signing anyway, reprepro is no suitable tool
for creating a full mirror that can be authenticated without adding the key of this
repository.
</dd>
<dt>Placing your files on your own</dt><dd>
Reprepro does all the calculation of filenames to save files as,
bookkeeping what files are there and what are needed and so on.
This cannot be switched off or disabled.
You can place files where reprepro will expect them and reprepro will use
them if their md5sum matches.
But reprepro is not suited if you want those files outside of a pool or in
places reprepro does not consider their canonical ones.
</dd>
<dt>Having different files with the same name</dt><dd>
take a look in the <a href="http://git.debian.org/?p=mirrorer/reprepro.git;a=blob_plain;f=docs/FAQ;hb=HEAD">FAQ</a> (currently question 1.2) why and how to avoid the problem.

</dd>
</dl>
</body>
</html>
reprepro 4.13.1-1build1 / usr / share / doc / reprepro / manual.html
