Table of Contents
List of Examples
ucr set
ucr get
is_ucr_true
ucr unset
ucr shell
nmap
for firewall port testingThis developer guide provides information to extend Univention Corporate Server. It it targeted at third party vendors who intend to provide applications for the Univention App Center and for power users who wish to deploy locally built or modified software.
Feedback is very welcome! Please either file a bug (see Appendix A) or send an
e-mail to <feedback@univention.de>
.
This chapter describes how software for UCS is packaged. For more details on packaging software in the Debian format, see Appendix B
UCS is based on the Debian distribution, which is using the deb format to package software.
The program dpkg
is used for handling a set of packages.
On installation packages are unpacked and configured, while on un-installation packages are de-configured and the files belonging to the packages are removed from the system.
On top of that the apt-tools provide a software repository, which allows software to be downloaded from central file servers.
Package
files provide an index of all packages contained in the repository, which is used to resolve dependencies between packages:
while dpkg
works on a set of packages given on the command line, apt-get
builds that set of packages and their dependencies before invoking dpkg
on this set.
apt-get
is a command line tool, which is fully described in its manual page apt-get(8).
A more modern version with a text based user interface is aptitude
, while synaptic
provides a graphical frontend.
On UCS systems the administrator is not supposed to use these tools directly. Instead all software maintenance can be done through the UMC, which maps the requests to invocations of the commands given above.
This chapter describes some simple examples using existing packages as examples. For downloading and building them, some packages must be installed on the system used as a development system. git is used to checkout the source files belonging to the packages. build-essential must be installed for building the package. devscripts provides some useful tools for maintaining packages.
This can be achieved by running the following command:
sudo apt-get install git build-essential devscripts
Source code: doc/developer-reference/packaging/testdeb/
Procedure 1.1. Checking out and building a UCS package
Create the top level working directory
mkdir work cd work/
Either fetch the latest source code from the GIT version control system or download the source code of the currently packaged version.
Checkout example package from GIT version control system
git checkout https://github.com/univention/univention-corporate-server.git cd univention-corporate-server/base/univention-ssh
Fetch the source code from the Univention Repository server
Enable unmaintained and source repository once
sudo ucr set repository/online/unmaintained=yes \ repository/online/sources=yes sudo apt-get update
Fetch source code
apt-get source univention-ssh cd univention-ssh-*/
Increment the version number of package to define a newer package
debchange --local work 'Private package rebuild'
Install the required build dependencies
sudo apt-get build-dep .
Build the binary package
dpkg-buildpackage -uc -us -b -rfakeroot
Locally install the new binary package
sudo dpkg -i ../univention-ssh_*_*.deb
The following example provides a walk-through for packaging a Python script called testdeb.py
.
It creates a file testdeb-
in the DATE
-time
/tmp/
directory.
A directory needs to be crated for each source package, which hosts all other files and sub-directories.
mkdir testdeb-0.1 cd testdeb-0.1
The file testdeb.py
, which is the program to be installed, will be put into that directory.
#!/usr/bin/env python """ Example for creating UCS packages. """ import time now = time.localtime() filename = '/tmp/testdeb-%s' % time.strftime('%y%m%d%H%M', now) tmpfile = open(filename, 'a') tmpfile.close()
In addition to the files to be installed some meta-data needs to be created in the debian/
sub-directory.
This directory contains several files, which are needed to build a Debian package.
The files and their format will be described in the following sections.
To create an initial debian/
directory with all template files, invoke the dh_make(8) command provided by the package dh-make:
dh_make --native --single --email user@example.com
Here several options are given to create the files for a source package, which contains all files in one archive and only creates one binary package at the end of the build process. More details are given in Section B.2.
The program will output the following information:
Maintainer name : John Doe Email-Address : user@example.com Date : Thu, 28 Feb 2013 08:11:30 +0100 Package Name : testdeb Version : 0.1 License : blank Type of Package : Single Hit <enter> to confirm:
The package name testdeb and version “0.1” were determined from the name of the directory testdeb-0.1
, the maintainer name and address were gathered from the UNIX account information.
After pressing the enter key some warning message will be shown:
Currently there is no top level Makefile. This may require additional tuning. Done. Please edit the files in the debian/ subdirectory now. You should also check that the testdeb Makefiles install into $DESTDIR and not in / .
Since this example is created from scratch, the missing Makefile
is normal and this warning can be ignored.
Instead of writing a Makefile to install the single executable, dh_install
will be used later to install the file.
Since the command completed successfully, several files were created in the debian/
directory.
Most of them are template files, which are unused in this example.
To improve understandability they are deleted:
rm debian/*.ex debian/*.EX rm debian/README* debian/doc
The remaining files are required and control the build process of all binary packages. Most of them don't need to be modified for this example, but others must be completed using an editor.
debian/control
The file contains general information about the source and binary packages. It needs to be modified to include a description and contain the right build dependencies:
Source: testdeb Section: univention Priority: optional Maintainer: John Doe <user@example.com> Build-Depends: debhelper (>= 7) Standards-Version: 3.7.2 Package: testdeb Architecture: all Depends: ${misc:Depends} Description: An example package for the developer guide This purpose of this package is to describe the structure of a Debian packages. It also documents . * the structure of a Debian/Univention package * installation process. * content of packages * format and function of control files . For more information about UCS, refer to: http://www.univention.de/
debian/rules
This file has a Makefile syntax and controls the package build process. Because there is no special handling needed in this example, the default file can be used unmodified.
#!/usr/bin/make -f %: dh $@
Note that tabulators must be used for indention in this file.
debian/testdeb.install
To compensate the missing Makefile
, dh_install(1) is used to install the executable.
dh_install
is indirectly called by dh
from the debian/rules
file.
To install the program into /usr/bin/
, the file needs to be created manually containing the following single line:
testdeb.py usr/bin/
Note that the path is not absolute but relative.
debian/testdeb.postinst
Since for this example the program should be invoked automatically during package installation, this file needs to be crated. In addition to just invoking the program shipped with the package itself, it also shows how Univention Configuration Registry variables can be set (see Section 2.1.1):
#! /bin/sh set -e case "$1" in configure) # invoke sample program testdeb.py # Set UCR variable if previously unset ucr set repository/online/server?https://updates.software-univention.de/ # Force UCR variable on upgrade from previous package only if dpkg --compare-versions "$2" lt-nl 0.1-2 then ucr set timeserver1=time.fu-berlin.de fi ;; abort-upgrade|abort-remove|abort-deconfigure) ;; *) echo "postinst called with unknown argument \`$1'" >&2 exit 1 ;; esac #DEBHELPER# exit 0
debian/changelog
The file is used to keep track of changes done to the packaging. For this example the file should look like this:
testdeb (0.1-1) unstable; urgency=low * Initial Release. -- John Doe <user@example.com> Mon, 21 Mar 2013 13:46:39 +0100
debian/copyright
This file is used to collect copyright related information. It is critical for Debian only, which need this information to guarantee that the package is freely redistributable. For this example the file remains unchanged.
The copyright
and changelog
file are installed to the /usr/share/doc/testdeb/
directory on the target system.
debian/compat
,debian/source/format
These files control some internal aspects of the package build process. They can be ignored for the moment and are further described in Section B.3.
Now the package is ready and can be built by invoking the following command:
dpkg-buildpackage -us -uc
The command should then produce the following output:
dpkg-buildpackage: source package testdeb dpkg-buildpackage: source version 0.1-1 dpkg-buildpackage: source changed by John Doe <user@example.com> dpkg-buildpackage: host architecture amd64 dpkg-source --before-build testdeb fakeroot debian/rules clean dh clean dh_testdir dh_auto_clean dh_clean dpkg-source -b testdeb dpkg-source: Information: Quellformat »3.0 (native)« wird verwendet dpkg-source: Information: testdeb wird in testdeb_0.1-1.tar.gz gebaut dpkg-source: Information: testdeb wird in testdeb_0.1-1.dsc gebaut debian/rules build dh build dh_testdir dh_auto_configure dh_auto_build dh_auto_test fakeroot debian/rules binary dh binary dh_testroot dh_prep dh_installdirs dh_auto_install dh_install dh_installdocs dh_installchangelogs dh_installexamples dh_installman dh_installcatalogs dh_installcron dh_installdebconf dh_installemacsen dh_installifupdown dh_installinfo dh_pysupport dh_pysupport: This program is deprecated, you should use dh_python2 instead. Migration guide: http://deb.li/dhs2p dh_installinit dh_installmenu dh_installmime dh_installmodules dh_installlogcheck dh_installlogrotate dh_installpam dh_installppp dh_installudev dh_installwm dh_installxfonts dh_installgsettings dh_bugfiles dh_ucf dh_lintian dh_gconf dh_icons dh_perl dh_usrlocal dh_link dh_compress dh_fixperms dh_installdeb dh_gencontrol dh_md5sums dh_builddeb dpkg-deb: building package `testdeb' in `../testdeb_0.1-1_all.deb'. dpkg-genchanges -b >../testdeb_0.1-1_amd64.changes dpkg-genchanges: binary-only upload - not including any source code dpkg-source --after-build testdeb dpkg-buildpackage: full upload; Debian-native package (full source is included)
The binary package file testdeb_0.1-1_all.deb
is stored in the parent directory.
When it is installed manually using dpkg -i ../testdeb_0.1-2_all.deb
as root, the Python script is installed as /usr/bin/testdeb.py
.
It is automatically invoked by the postint
script, so a file named /tmp/testdeb-
has been created, too.
date
-time
Congratulations! You've successfully built your first own Debian package.
Until now the binary package is only available locally, thus for installation it needs to be copied manually to each host and must be installed manually using dpkg -i
.
If the package required additional dependencies, the installation process will abort, since packages are not downloaded by dpkg
, but by apt
.
To support automatic installation and dependency resolution, the package must be put into an apt repository, which needs to be made available through http
or some other mechanism.
For this example the repository is created below /var/www/repository/
, which is exported by default on all UCS systems, where apache2 is installed.
Below that directory several other sub-directories and files must be created to be compatible with the UCS Updater.
The following example commands create a repository for UCS version 4.1 with the component name testcomp
:
WWW_BASE="/var/www/repository/4.1/maintained/component" TESTCOMP="testcomp/all" install -m755 -d "$WWW_BASE/$TESTCOMP" install -m644 -t "$WWW_BASE/$TESTCOMP" *.deb ( cd "$WWW_BASE" && rm -f "$TESTCOMP/Packages"* && apt-ftparchive packages "$TESTCOMP" > "Packages" && gzip -9 < "Packages" > "$TESTCOMP/Packages.gz" && mv "Packages" "$TESTCOMP/Packages" )
This repository can be included on any UCS system by appending the following line to /etc/apt/sources.list
, assuming the FQDN of the host providing the repository is named repository.server
:
deb http://repository.server/repository/4.1/maintained/component testcomp/all/
It is important that the directory, from were the apt-ftparchive
command is invoked, matches the first string given in the sources.list
file after the deb
prefix.
The URL together with the suffix testcomp/all/
not only specifies the location of the Packages
file, but is also used as the base URL for all packages listed in the Packages
file.
Instead of editing the sources.list
file directly, the repository can also be included as a component, which can be configured by setting several UCR variables.
As UCR variables can also be configured through UDM policies, this simplifies the task of installing packages from such a repository on may hosts.
For the repository above the following variables need to be set:
ucr set \ repository/online/component/testcomp=yes \ repository/online/component/testcomp/server=repository.server \ repository/online/component/testcomp/prefix=repository
The openSUSE Build Service (OBS) is a framework to generate packages for a wide range of distributions. Additional information can be found at https://build.opensuse.org/. If OBS is already used to build packages for other distributions, it can also be used for Univention Corporate Server builds. The build target for UCS 4.1 is called Univention UCS 4.1. Note that OBS doesn't handle the integration steps described in later chapters (e.g. the use of Univention Configuration Registry templates).
The Univention Config Registry (UCR) is a local mechanism, which is used on all UCS system roles to consistently configure all services and applications. It consists of a database, were the currently configured values are stored, and a mechanism to trigger certain actions, when values are changed. This is mostly used to create configuration files from templates by filling in the configured values. In addition to using simple place holders its also possible to use Python code for more advanced templates or to call external programs when values are changed. UCR values can also be configured through an UDM policy in Univention directory service (LDAP), which allows values to be set consistently for multiple hosts of a domain.
Univention Configuration Registry provides two interfaces, which allows easy access from shell scripts and Python programs.
univention-config-registry
(and its alias ucr
) can be invoked directly from shell.
The most commonly used functions are:
ucr
set [
key
=value
] | [
key
?value
]...
Set Univention Configuration Registry variable key
to the given value
.
Using =
forces an assignment, while ?
only sets the value if the variable is unset.
ucr
get
key
Return the current value of the Univention Configuration Registry variable key
.
Example 2.2. Use of ucr get
case "$(ucr get system/role)" in domaincontroller_*) echo "Running on a UCS Domain Controller" ;; esac
For variables containing boolean values the shell-library-function is_ucr_true
from key
/usr/share/univention-lib/ucr.sh
should be used.
It returns 0 (success) for the values "1", "yes", "on", "true", "enable", "enabled", 1 for the negated values "0", "no", "off", "false", "disable", "disabled".
For all other values it returns a value of 2 to indicate inappropriate usage.
Example 2.3. Use of is_ucr_true
. /usr/share/univention-lib/ucr.sh if is_ucr_true repository/online/unmaintained then echo "Unmaintained is enabled" fi
ucr
unset
key
...
Unset the Univention Configuration Registry variable key
.
ucr
shell [
key
...]
Export some or all Univention Configuration Registry variables in a shell compatible manner as environment variables.
All shell-incompatible characters in variable names are substituted by underscores (_
).
Example 2.5. Use of ucr shell
eval "$(ucr shell)" case "$server_role" in domaincontroller_*) echo "Running on a UCS Domain Controller serving $ldap_base" ;; esac
It is often easier to export all variables once and than reference the values through shell variables.
Be careful with shell quoting, since several Univention Configuration Registry variables contain shell meta characters.
Use eval "$(ucr shell)"
.
ucr
is installed as /usr/sbin/ucr
, which is not on the search path $PATH
of normal users.
Changing variables requires root access to /etc/univention/base.conf
, but reading works for normal users too, if /usr/sbin/ucr
is invoked directly.
UCR also provides a Python binding, which can be used from any Python program.
An instance of univention.config_registry.ConfigRegistry
needs to be crated first.
After loading the current database state with load()
the values can be accessed by using the instance like a Python dictionary:
Example 2.6. Reading a Univention Configuration Registry variable in Python
from univention.config_registry import ConfigRegistry ucr = ConfigRegistry() ucr.load() print ucr['variable/name'] print ucr.get('variable/name', '<not set>')
For variables containing boolean values the methods is_true()
and is_false()
should be used.
The former returns True
for the values "1", "yes", "on", "true", "enable", "enabled", while the later one returns True
for the negated values "0", "no", "off", "false", "disable", "disabled".
Both methods accept an optional argument default
, which is returned as-is when the variable is not set.
Example 2.7. Reading boolean Univention Configuration Registry variables in Python
if ucr.is_true('repository/online/unmaintained'): print "unmaintained is explicitly enabled" if ucr.is_true('repository/online/unmaintained', True): print "unmaintained is enabled" if ucr.is_false('repository/online/unmaintained'): print "unmaintained is explicitly disabled" if ucr.is_false('repository/online/unmaintained', True): print "unmaintained is disabled"
Modifying variables requires a different approach.
The function ucr_update()
should be used to set and unset variables.
Example 2.8. Changing Univention Configuration Registry variables in Python
from univention.config_registry.frontend import ucr_update ucr_update(ucr, { 'foo': 'bar', 'baz': '42', 'bar': None, })
The function ucr_update()
requires an instance of ConfigRegistry
as its first argument.
The method is guaranteed to be atomic and internally uses file locking to prevent race conditions.
The second argument must be a Python dictionary mapping UCR variable names to their new value.
The value must be either a string or None
, which is used to unset the variable.
As an alternative the old functions handler_set()
and handler_unset()
can still be used to set and unset variables.
Both functions expect an array of strings with the same syntax as used with the command line tool ucr
.
As the functions handler_set()
and handler_unset()
don't automatically update any instance of ConfigRegistry
, the method load()
has to be called manually afterwards to reflect the updated values.
Example 2.9. Setting and unsetting Univention Configuration Registry variables in Python
from univention.config_registry import handler_set, handler_unset handler_set(['foo=bar', 'baz?42']) handler_unset(['foo', 'bar'])
Packages can use the UCR functionality to create customized configuration files themselves. UCR diverts files shipped by Debian packages and replaces them by generated files. If variables are changed, the affected files are committed, which regenerated their content. This diversion is persistent and even outlives updates, so they are not overwritten by configuration files of new packages.
For this, packages need to ship additional files:
conffiles/path/to/file
debian/package
.univention-config-registry
debian/package
.univention-config-registry-variables
debian/package
.univention-config-registry-categories
debian/package
.univention-config-registry-services
In addition to these files code needs to be inserted into the package maintainer scripts (see Section B.3.5), which registers and unregisters these files.
This is done by calling univention-install-config-registry
from debian/rules
during the package build binary
phase.
The command is part of the univention-config-dev package, which needs to be added as a Build-Depends
build dependency of the source package in debian/control
.
debian/package
.univention-config-registry
This file describes all template files in the package.
The file is processed and copied by univention-install-config-registry
into /etc/univention/templates/info/
when the package is built.
It can consist of multiple sections, where sections are separated by one blank line. Each section consists of multiple key-value-pairs separated by a colon followed by one blank. A typical entry has the following structure:
Type: <type> [Multifile|File]: <filename> [Subfile: <fragment-filename>] Variables: <variable1> ...
Type
specifies the type of the template, which the following sections describe in more detail.
File
A single file template is specified as type file
.
It defines a template, were the target file is created from only a single source file.
A typical entry hat the following structure:
Type: file File: <filename> Variables: <variable1> User: <owner> Group: <group> Mode: <file-mode> Preinst: <module> Postinst: <module> ...
The following keys can be used:
File
(required)
Specifies both the target and source file name, which are identical.
The source file containing the template must be put below the conffiles/
directory.
The file can contain any textual content and is processed as described in Section 2.3.
The template file is installed to /etc/univention/templates/files/
.
Variables
(optional)
This key can be given multiple times and specifies the name of UCR variables, which trigger the file commit process.
This is normally only required for templates using @!@
Python code regions.
Variables used in @%@
sections do not need to be listed explicitly, since ucr
extracts them automatically.
The variable name is actually a Python regular expression, which can be used to match, for example, all variable names starting with a common prefix.
User
(optional),Group
(optional),Mode
(optional)root:root
is used by default and the file mode is inherited from the source template.
Preinst
(optional),Postinst
(optional)
These specify the name of a Python module located in /etc/univention/templates/modules/
, which is called before and after the target file is re-created.
The module must implement the following two functions:
def preinst(config_registry, changes): pass def postinst(config_registry, changes): pass
Each function receives two arguments:
The first argument config_registry
is a reference to an instance of ConfigRegistry
.
The second argument changes
is a dictionary of 2-tuples, which maps the names of all changed variables to (old-value
, new-value
).
univention-install-config-registry
installs the module file to /etc/univention/templates/modules/
.
If a script /etc/univention/templates/scripts/
exists, it will be called after the file is committed.
The script is called with the argument full-path-to-file
postinst
.
It receives the list of changed variables as documented in Section 2.2.1.3.
Multifile
A multi file template is specified once as type multifile
, which describes the target file name.
In addition to that multiple sections of type subfile
are used to describe source file fragments, which are concatenated to form the final target file.
A typical multifile has the following structure:
Type: multifile Multifile: <target-filename> User: <owner> Group: <group> Mode: <file-mode> Preinst: <module> Postinst: <module> Variables: <variable1> Type: subfile Multifile: <target-filename> Subfile: <fragment-filename> Variables: <variable1> ...
The following keys can be used:
Multifile
(required)multifile
entry to its corresponding subfile
entries.
Subfile
(required)
The source file containing the template fragment must be put below the conffiles/
directory in the Debian source package.
The file can contain any textual content and is processed as described in Section 2.3.
The template file is installed to /etc/univention/templates/files/
.
Common best practice is to start the filename with two digits to allow consistent sorting and to put the file in the directory named like the target filename suffixed by .d
, that is conffiles/
.
target-filename
.d/00fragment-filename
Variables
(optional)multifile
and subfile
sections.
The variables from all sections trigger the commit of the target file.
Until UCS-2.4 only the multifile
section was used, since UCS-3.0 the subfile
section should be preferred (if needed).
User
(optional),Group
(optional),Mode
(optional),Preinst
(optional),Postinst
(optional)file
.
The same script hook as above for file
is also supported.
Script
A script template allows an external program to be called when specific UCR variables are changed. A typical script entry has the following structure:
Type: script Script: <filename> Variables: <variable1>
The following keys can be used:
Script
(required)
Specifies the filename of an executable, which is installed to /etc/univention/templates/scripts/
.
The script is called with the argument generate
.
It receives the list of changed variables on standard input.
For each changed variable a line containing the name of the variable, the old value, and the new value separated by @%@
is sent.
Variables
(required)Module
A module template allows a Python module to be run when specific UCR variables are changed. A typical module entry has the following structure:
Type: module Module: <filename> Variables: <variable1>
The following keys can be used:
Module
(required)
Specifies the filename of a Python module, which is installed to /etc/univention/templates/modules/
.
The module must implement the following function:
def handler(config_registry, changes): pass
The function receives two arguments:
The first argument config_registry
is a reference to an instance of ConfigRegistry
.
The second argument changes
is a dictionary of 2-tuples, which maps the names of all changed variables to (old-value
, new-value
).
univention-install-config-registry
installs the module to /etc/univention/templates/modules/
.
Variables
(required)debian/package
.univention-config-registry-variables
For UCR variables a description should be registered.
This description is shown in the Univention Config Registry module of the UMC as a mouse-over.
It can also be queried by running ucr info
on the command line.
variable/name
The description is provided on a per-package basis as a file, which uses the ini-style format.
The file is processed and copied by univention-install-config-registry-info
into /etc/univention/registry.info/variables/
.
The command univention-install-config-registry-info
is invoked indirectly by univention-install-config-registry
, which should be called instead from debian/rules
.
For each variable a section of the following structure is defined:
[<variable/name>] Description[en]=<description> Description[<language>]=<description> Type=<type> ReadOnly=<yes|no> Categories=<category,...>
[
variable/name
]
(required)For each variable description one section needs to be created. The name of the section must match the variable name.
To describe multiple variables with a common prefix and/or suffix, the regular expression .*
can be used to match any sequence of characters.
This is the only supported regular expression!
Description[
language
]
(required)Type
(required)str
for strings, bool
for boolean values, and int
for integers.
ReadOnly
(optional)ucr
.
Valid values are true
for read-only and false
, which is the default.
Categories
(required)debian/package
.univention-config-registry-categories
UCR variables can be grouped into categories, which can help administrators to find related settings.
Categories are referenced from .univention-config-registry-variables
files (see Section 2.2.2).
They are created on-the-fly, but can be described further by explicitly defining them in a .univention-config-registry-categories
file.
The description is provided on a per-package basis as a file, which uses the ini-style format.
The file is processed and copied by univention-install-config-registry-info
into /etc/univention/registry.info/categories/
.
The command univention-install-config-registry-info
is invoked indirectly by univention-install-config-registry
, which should be called instead from debian/rules
.
For each category a section of the following structure is defined:
[<category-name>] name[en]=<name> name[<language>]=<translated-name> icon=<file-name>
[
category-name
]
name[
language
]
(required)icon
(required)debian/package
.univention-config-registry-services
Long running services should be registered with UCR and UMC. This enables administrators to control these daemons using the UMC module System services.
The description is provided on a per-package basis as a file, which uses the ini-style format.
The file is processed and copied by univention-install-service-info
into /etc/univention/service.info/services/
.
The command univention-install-service-info
is invoked indirectly by univention-install-config-registry
, which should be called instead from debian/rules
.
For each service a section of the following structure is defined:
[<service-name>] description[<language>]=<description> start_type=<service-name>/autostart systemd=<service-name>.service icon=<service/icon_name> programs=<executable>
[
service-name
]
/etc/init.d/
.
description[
language
]
(required)start_type
(required)/usr/share/univention-config-registry/init-autostart.lib
to evaluate the setting from the init-script of the service.
If the variable is set to false
or no
, the service should never be started.
If the variable is set to manually
, the service should not be started automatically, but invoking the init-script directly with start
should still start the service.
systemd
(optional)start_type
is used.
This defaults to the name of the service plus the suffix .service
.
programs
(required)/proc/*/cmdline
.
To check the processes for additional arguments, the command can also consist of additional shell-escaped arguments.
icon
(optional)conffiles/path/to/file
For each file, which should be written, one or more template files need be to created below the conffiles/
directory.
For a single-File template (see Section 2.2.1.1), the filename must match the filename given in the File:
stanza of the file entry itself.
For a Multifile template (see Section 2.2.1.2), the filename must match the filename given in the File:
stanza of the subfile entries.
Each template file is normally a text file, where certain sections get substituted by computed values during the file commit. Each section starts and ends with a special marker. UCR currently supports the following kinds of markers:
@%@
variable reference
Sections enclosed in @%@
are simple references to Univention Configuration Registry variable.
The section is replaced inline by the current value of the variable.
If the variable is unset, an empty string is used.
ucr
scans all file
s and subfile
s on registration.
All Univention Configuration Registry variables used in @%@
are automatically extracted and registered for triggering the template mechanism.
They don't need to be explicitly enumerated with Variables:
-statements in the file debian/
.
package
.univention-config-registry
@!@
Python code
Sections enclosed in @!@
contain Python code.
Everything printed to STDOUT by these sections is inserted into the generated file.
The Python code can access the configRegistry
[1] variable, which is an already loaded instance of ConfigRegistry
.
Each section is evaluated separately, so no state is kept between different Python sections.
All Univention Configuration Registry variables used in a @!@
Python section must be manually matched by a Variables:
statement in the debian/
file.
Otherwise the file is not updated on changes of the UCR variable.
package
.univention-config-registry
@%@UCRWARNING=%
PREFIX
@%@
,@%@UCRWARNING_ASCII=%
PREFIX
@%@
This variant of the variable reference inserts a warning text, which looks like this:
# Warning: This file is auto-generated and might be overwritten by # univention-config-registry. # Please edit the following file(s) instead: # Warnung: Diese Datei wurde automatisch generiert und kann durch # univention-config-registry überschrieben werden. # Bitte bearbeiten Sie an Stelle dessen die folgende(n) Datei(en): # # /etc/univention/templates/files/etc/hosts.d/00-base # /etc/univention/templates/files/etc/hosts.d/20-static # /etc/univention/templates/files/etc/hosts.d/90-ipv6defaults #
It should be inserted once at the top to prevent the user from editing the generated file. For single File templates, it should be on the top of the template file itself. For Multifile templates, it should only be on the top the first subfile.
Everything between the equal sign and the closing @%@
defines the PREFIX
, which is inserted at the beginning of each line of the warning text.
For shell scripts, this should be #
, but other files use different characters to start a comment.
For files, which don't allow comments, the header should be skipped.
Several file formats require the file to start with some magic data.
For example shell scripts must start with a hash-bang (#!
) and XML files must start with <?xml version="1.0" encoding="UTF-8"?>
(if used).
Make sure to put the warning after these headers!
The UCRWARNING_ASCII
variant only emits 7-bit ASCII characters, which can be used for files, which are not 8 bit clean or unicode aware.
During package build time univention-install-config-registry
needs to be called.
This should be done by overriding the dh_auto_install_target
in debian/rules
:
override_dh_auto_install: univention-install-config-registry dh_auto_install
This invocation copies the referenced files to the right location in the binary package staging area debian/
.
Internally package
/etc/univention/univention-install-config-registry-info
and univention-install-service-info
are invoked, which should not be called explicitly anymore.
The calls also insert code into the files debian/
, package
.preinst.debhelperdebian/
and package
.postinst.debhelperdebian/
to register and un-register the templates.
Therefore it's important that customized maintainer scripts use the package
.prerm.debhelper#DEBHELPER#
marker, so that the generated code gets inserted into the corresponding preinst
, postinst
and prerm
files of the generated binary package.
The invocation also adds univention-config to misc:Depends
to ensure that the package is available during package configuration time.
Therefore it's important that ${misc:Depends}
is used in the Depends
line of the package section in the debian/control
file.
Package: ... Depends: ..., ${misc:Depends}, ...
This sections contains several simple examples for the use of Univention Configuration Registry. The complete source of these examples is available separately. The download location is given in each example below. Since almost all Univention Corporate Server packages use UCR, their source code provides additional examples.
This example provides a template for /etc/papersize
, which is used to configure the default paper size.
A Univention Configuration Registry variable print/papersize
is registered, which can be used to configure the papersize.
Source code: doc/developer-reference/ucr/papersize/
conffiles/etc/papersize
The template file only contains one line. Please note that this file does not start with the “UCRWARNING”, since the file must only contain the paper size and no comments.
@%@print/papersize@%@
debian/papersize.univention-config-registry
The file defines the templates and is processed by univention-install-config-registry
during the package build and afterwards by univention-config-registry
during normal usage.
Type: file File: etc/papersize
debian/papersize.univention-config-registry-variables
The file describes the newly defined Univention Configuration Registry variable.
[print/papersize] Description[en]=specify preferred paper size [a4] Description[de]=Legt die bevorzugte Papiergröße fest [a4] Type=str Categories=service-cups
debian/papersize.postinst
Sets the Univention Configuration Registry variable to a default value after package installation.
#!/bin/sh #DEBHELPER# ucr set print/papersize?a4 exit 0
debian/rules
Invoke univention-install-config-registry
during package build to install the files to the appropriate location.
It also creates the required commands for the maintainer scripts (see Section B.3.5) to register and unregister the templates during package installation and removal.
#!/usr/bin/make -f override_dh_auto_install: dh_auto_install univention-install-config-registry %: dh $@
Note that tabulators must be used for indention in this Makefile-type file.
debian/control
The automatically generated dependency on univention-config is inserted by univention-install-config-registry
via debian/papersize.substvars
.
Source: papersize Section: univention Priority: optional Maintainer: Univention GmbH <packages@univention.de> Build-Depends: debhelper (>= 7), univention-config-dev, Standards-Version: 3.7.2 Package: papersize Architecture: all Depends: ${misc:Depends} Description: An example package to configure the papersize This purpose of this package is to show how Univention Config Registry is used. . For more information about UCS, refer to: http://www.univention.de/
This example provides templates for /etc/hosts.allow
and /etc/hosts.deny
, which is used to control access to system services.
See hosts_access(5) for more details.
Source code: doc/developer-reference/ucr/hosts/
conffiles/etc/hosts.allow.d/00header
,conffiles/etc/hosts.deny.d/00header
The first file fragment of the file.
It starts with @%@UCRWARNING=# @%@
, which is replaced by the warning text and a list of all subfiles.
@%@UCRWARNING=# @%@ # /etc/hosts.allow: list of hosts that are allowed to access the system. # See the manual pages hosts_access(5) and hosts_options(5).
conffiles/etc/hosts.allow.d/50dynamic
,conffiles/etc/hosts.deny.d/50dynamic
A second file fragment, which uses Python code to insert access control entries configured through the Univention Configuration Registry variables hosts/allow/
and hosts/deny/
.
@!@ for key, value in sorted(configRegistry.items()): if key.startswith('hosts/allow/'): print value @!@
debian/hosts.univention-config-registry
The file defines the templates and is processed by univention-install-config-registry
.
Type: multifile Multifile: etc/hosts.allow Type: subfile Multifile: etc/hosts.allow Subfile: etc/hosts.allow.d/00header Type: subfile Multifile: etc/hosts.allow Subfile: etc/hosts.allow.d/50dynamic Variables: ^hosts/allow/.* Type: multifile Multifile: etc/hosts.deny Type: subfile Multifile: etc/hosts.deny Subfile: etc/hosts.deny.d/00header Type: subfile Multifile: etc/hosts.deny Subfile: etc/hosts.deny.d/50dynamic Variables: ^hosts/deny/.*
debian/hosts.univention-config-registry-variables
The file describes the newly defined Univention Configuration Registry variables.
[hosts/allow/.*] Description[en]=An permissive access control entry for system services, e.g. "ALL: LOCAL" Description[de]=Eine erlaubende Zugriffsregel für Systemdienste, z.B. "ALL: LOCAL". Type=str Categories=service-net [hosts/deny/.*] Description[en]=An denying access control entry for system services, e.g. "ALL: ALL". Description[de]=Eine verbietende Zugriffsregel für Systemdienste, z.B. "ALL: ALL". Type=str Categories=service-net
This example provides a template to control the atd
service through an Univention Configuration Registry variable atd/autostart
.
Source code: doc/developer-reference/ucr/service/
conffiles/etc/init.d/atd
The template replaces the original file with a version, which checks the Univention Configuration Registry variable atd/autostart
before starting the at
daemon.
Please note that the “UCRWARNING” is put after the hash-bash line.
#! /bin/sh @%@UCRWARNING=# @%@ ### BEGIN INIT INFO # Provides: atd # Required-Start: $syslog $time $remote_fs # Required-Stop: $syslog $time $remote_fs # Default-Start: 2 3 4 5 # Default-Stop: 0 1 6 # Short-Description: Deferred execution scheduler # Description: Debian init script for the atd deferred executions # scheduler ### END INIT INFO # pidfile: /var/run/atd.pid # # Author: Ryan Murray <rmurray@debian.org> # PATH=/bin:/usr/bin:/sbin:/usr/sbin DAEMON=/usr/sbin/atd PIDFILE=/var/run/atd.pid test -x $DAEMON || exit 0 . /lib/lsb/init-functions case "$1" in start) # check ucr autostart setting IAL="/usr/share/univention-config-registry/init-autostart.lib" if [ -f "$IAL" ]; then . "$IAL" check_autostart atd atd/autostart fi log_daemon_msg "Starting deferred execution scheduler" "atd" start_daemon -p $PIDFILE $DAEMON log_end_msg $? ;; stop) log_daemon_msg "Stopping deferred execution scheduler" "atd" killproc -p $PIDFILE $DAEMON log_end_msg $? ;; force-reload|restart) $0 stop $0 start ;; status) status_of_proc -p $PIDFILE $DAEMON atd && exit 0 || exit $? ;; *) echo "Usage: $0 {start|stop|restart|force-reload|status}" exit 1 ;; esac exit 0
Note the inclusion of init-autostart.lib
and use of check_autostart
.
debian/service.univention-config-registry
The file defines the templates.
Type: file File: etc/init.d/atd Mode: 755 Variables: atd/autostart
Note the additional Mode
statement to mark the file as executable.
debian/service.univention-config-registry-variables
The file adds a description for the Univention Configuration Registry variable atd/autostart
.
[atd/autostart] Description[en]=Automatically start the AT daemon on system startup [yes] Description[de]=Automatischer Start des AT-Dienstes beim Systemstart [yes] Type=bool Categories=service-at
debian/service.postinst
Set the Univention Configuration Registry variable to automatically start the atd
on new installations.
#!/bin/sh #DEBHELPER# ucr set atd/autostart?yes exit 0
debian/control
univention-base-files must be added manually as an additional dependency, since it is used from within the shell code.
Source: service Section: univention Priority: optional Maintainer: Univention GmbH <packages@univention.de> Build-Depends: debhelper (>= 7), univention-config-dev, Standards-Version: 3.7.2 Package: service Architecture: all Depends: ${misc:Depends}, univention-base-files, Description: An example package to configure services This purpose of this package is to show how Univention Config Registry is used. . For more information about UCS, refer to: http://www.univention.de/
[1]
Historically Univention Configuration Registry was named “Univention Base Config”.
For backward compatibility the alias baseConfig
is still provided.
It should not be used anymore and will be removed in a future version of UCS.
An UCS system is normally joined into a domain. This establishes a trust relation between the different hosts, which enables users to access services provided by any host of the domain.
Joining a system into a domain requires write permission to create and modify entries in the Univention directory service (LDAP).
Local root
permission on the joining host is not sufficient to get write access to the domain wide LDAP service.
Instead valid LDAP credentials must be entered interactively by the administrator doing the join.
Packages requiring write access to the Univention directory service can provide so called join scripts.
They are installed into /usr/lib/univention-install/
.
The name of each join script is normally derived from the name of the binary package containing it.
It is prefixed with a two-digit number, which is used to order the scripts lexicographically.
The filename either ends in .inst
or .uinst
, which distinguishes between join script and unjoin script (see Section 3.5).
The file must have the executable permission bits set.
For each join script a version number is tracked. This is used to skip re-executing join scripts, which already have been executed. This is mostly a performance optimization, but is also used to find join scripts which need to be run.
The text file /var/univention-join/status
is used to keep track of the state of all join scripts.
For each successful run of a join script a line is appended to that file.
That record consists of three space separated entries:
$script_name v$version successful
.inst
suffix, usually corresponding to the package name.
v
.
It is used to keep track of the latest version of the join script, which has been run successfully.
This is used to identify, which join scripts need to be executed and which can be skipped, because they were already executed in the past.
If a new version of the join script is invoked, it just appends a new record with a higher version number at the end of the file.
There exist three commands related to running join scripts:
univention-join
When univention-join
is invoked, a machine account is created.
The distinguished name (dn) of that entry is stored locally in the Univention Configuration Registry variable ldap/hostdn
.
A random password is generated, which is stored in the file /etc/machine.secret
.
After that the file /var/univention-join/status
is cleared and all join scripts located in /usr/lib/univention-install/
are executed in lexicographical order.
univention-run-join-scripts
This command is similar to univention-join
, but skips the first step of creating a machine account.
Only those join scripts are executed, whose current version is not yet registered in /var/univention-join/status
.
univention-check-join-status
This command only checks for join scripts in /usr/lib/univention-install/
, whose version is not yet registered in /var/univention-join/status
.
When packages are installed, it depends on the server role, if join scripts are invoked automatically from the postinst
Debian maintainer script or not.
This only happens on master and backup domain controller system roles, where the local root
user has access to the file containing the LDAP credentials.
On all other system roles the join scripts need to be run manually by invoking univention-run-join-scripts
or doing so through UMC.
Similar to the Debian maintainer scripts (see Section B.3.5) they should be idem-potent: They should transform the system from any state into the state required by the package, that is:
Join scripts may be called from multiple system roles and different versions. Therefore it is important that these scripts do not destroy or remove data still used by other systems!
This example provides a template for writing join scripts. The package is called join-template and just contains a join and an unjoin script. They demonstrate some commonly used functions.
Source code: doc/developer-reference/join/join-template/
50join-template.inst
The join script in UCS packages is typically located in the package root directory. It has the following base structure:
#!/bin/sh VERSION=1 . /usr/share/univention-join/joinscripthelper.lib joinscript_init SERVICE="MyService" eval "$(ucr shell)" . /usr/share/univention-lib/ldap.sh ucs_addServiceToLocalhost "$SERVICE" "$@" udm "computers/$server_role" modify "$@" \ --dn "$ldap_hostdn" \ --set reinstall=0 || die # create container for extended attributes to be placed in udm container/cn create "$@" \ --ignore_exists \ --position "cn=custom attributes,cn=univention,$ldap_base" \ --set name="myservice" || die # some extended attributes would be added here joinscript_save_current_version exit 0
Please note the essential argument "$@"
when udm
is invoked, which passes on the required LDAP credentials described in Section 4.4.
debian/join-template.install
The scripts need to be installed into /usr/lib/univention-install/
, which is achieved by the following lines:
50join-template.inst usr/lib/univention-install/ 50join-template-uninstall.uinst usr/lib/univention-uninstall/
Note that this package also installs an unjoin script.
debian/join-template.postinst
The join script should be invoked automatically on master and backup domain controller systems.
On all other system roles an administrator must run the join script manually through univention-run-join-scripts
.
#!/bin/sh #DEBHELPER# if [ "$1" = "configure" ] then uinst=/usr/lib/univention-install/50join-template-uninstall.uinst [ -e "$uinst" ] && rm "$uinst" fi . /usr/share/univention-lib/base.sh call_joinscript 50join-template.inst exit 0
debian/control
The package uses two shell libraries, which are described in more detail in Section 3.4.3. Both packages providing them must be added as additional runtime dependencies.
The unjoin functions were added to UCS 3.1-0 only as erratum update 81. Because of this the minimum versions must be specified explicitly.
Source: join-template Section: univention Priority: optional Maintainer: Univention GmbH <packages@univention.de> Build-Depends: debhelper (>= 7) Standards-Version: 3.7.2 Package: join-template Architecture: all Depends: univention-join (>= 5.0.20-1), shell-univention-lib (>= 2.0.17-1), ${misc:Depends} Description: An example package for join scripts This purpose of this package is to show how Univention Join scripts are used. . For more information about UCS, refer to: http://www.univention.de/
Join scripts must return the following exit codes:
0
The join script was successful and completed all tasks to join the software package on the system into the domain. All required entries in the Univention directory service were created or do already exist as expected.
The script will be marked as successfully run. As a consequence the join script will not be called again in this version.
1
The script did not complete and some task to fully join the system into the domain are still pending. Some entries could not be created in LDAP or exist in a state, which is incompatible with this version of the package.
The script needs to be run again after fixing the problem, either manually or automatically.
2
Some internal functions were called incorrectly. For example the credentials were wrong.
The script needs to be run again.
There exist two shell libraries, which provide functions which help in writing join scripts:
The package contains the shell library /usr/share/univention-join/joinscripthelper.lib
.
It provides functions related to updating the join status file.
joinscript_init
This function parses the status file and exits the shell script, if a record is found with a version greater or equal to value of the environment variable VERSION
.
The name of the join script is derived from $0
.
joinscript_save_current_version
This function appends a new record to the end of the status file using the version number stored in the environment variable VERSION
.
joinscript_check_any_version_executed
This function returns success (0
), if any previous version of the join scripts was successfully executed.
Otherwise it returns a failure (1
).
joinscript_check_specific_version_executed
version
This function returns success (0
), if the specified version version
of the join scripts was successfully executed.
Otherwise it returns a failure (1
).
joinscript_check_version_in_range_executed
min max
This function returns success (0
), if any successfully run version of the join script falls within the range min
..max
, inclusively.
Otherwise it returns a failure (1
).
joinscript_extern_init
join-script
The check commands mentioned above can also be used in other shell programs, which are not join scripts.
There the name of the join script to be checked must be explicitly given.
Instead of calling joinscript_init
, this function requires an additional argument specifying the name of the join-script
.
joinscript_remove_script_from_status_file
name
Removes the given join script from the join script status file /var/univention-join/status
.
The name
should be the basename of the joinscript without the prefixed digits and the suffix .inst
.
So if the joinscript /var/lib/univention-install/50join-template.inst
shall be removed, one has to execute joinscript_remove_script_from_status_file join-template
.
Primarily used in unjoin scripts.
die
A convenience function to exit the join script with an error code.
Used to guarantee that LDAP modifications were successful:
some_udm_create_call || die
These functions use the following environment variables:
VERSION
This variable must be set before joinscript_init
is invoked.
It specifies the version number of the join script and is used twice:
The version number should be incremented for a new version of the package, when the join script needs to perform additional modifications in LDAP compared to any previous packaged version.
The version number must be a positive integer.
The variable assignment in the join script must be on its own line.
It may optionally quote the version number with single quotes ('
) or double quotes ("
).
The following assignment are valid:
VERSION=1 VERSION='2' VERSION="3"
JS_LAST_EXECUTED_VERSION
This variable is initialized by joinscript_init
with the latest version found in the join status file.
If no version of the join script was ever executed and thus no record exists, the variable is set to 0
.
The join script can use this information to decide what to do on an upgrade.
The package contains the shell library /usr/share/univention-lib/base.sh
.
Since package version >= 2.0.17-1
it provides the following functions:
call_joinscript
[--binddn bind-dn
--bindpwd bind-password
] [XXjoin-script.inst
]
XXjoin-script.inst
from the directory /usr/lib/univention-install/
.
The optional LDAP credentials bind-dn
and bind-password
are passed on as-is.
call_joinscript_on_dcmaster
[--binddn bind-dn
--bindpwd bind-password
] [XXjoin-script.inst
]
call_joinscript
, but also checks the system role and only executes the script on the master domain controller.
remove_joinscript_status
[name
]
name
from the join script status file /var/univention-join/status
.
Note that this command does the same as joinscript_remove_script_from_status_file
provided by
univention-join (see Section 3.4.3.1).
call_unjoinscript
[--binddn bind-dn
--bindpwd bind-password
] [XXunjoin-script.uinst
]
unjoin-script
on master and backup domain controller systems.
The file name must be relative to the directory /usr/lib/univention-install/
.
The optional LDAP credentials bind-dn
and bind-password
are passed on as-is.
Afterwards the unjoin script is automatically deleted.
delete_unjoinscript
[XXunjoin-script.uinst
]
XXunjoin-script.uinst
if it does not belong to any package.
The file name must be relative to the directory /usr/lib/univention-install/
.
stop_udm_cli_server
univention-directory-manager
is used the first time a server is started automatically that caches some information about the available modules.
When changing some of this information (e.g. when adding or removing extended attributes) the server should be stopped manually.
The package also contains the shell library /usr/share/univention-lib/ldap.sh
.
It provides convenience functions to query the Univention directory service and modify objects.
For (un)join scripts the following functions might be important:
ucs_addServiceToLocalhost
servicename
[--binddn bind-dn
--bindpwd bind-password
]
Registers the additional service servicename
in the LDAP object representing the local host.
The optional LDAP credentials bind-dn
and bind-password
are passed on as-is.
ucs_removeServiceFromLocalhost
servicename
[--binddn bind-dn
--bindpwd bind-password
]
Removes the service servicename
from the LDAP object representing the local host, effectively reverting an ucs_addServiceToLocalhost
call.
The optional LDAP credentials bind-dn
and bind-password
are passed on as-is.
Example 3.2. Service unregistration in unjoin script
ucs_removeServiceFromLocalhost "MyService" "$@"
ucs_isServiceUnused
servicename
[--binddn bind-dn
--bindpwd bind-password
]
Returns 0
if no LDAP host object exists where the service servicename
is registered with.
Example 3.3. Check for unused service in unjoin script
if ucs_isServiceUnused "MyService" "$@" then uninstall_my_service fi
ucs_registerLDAPExtension
[--binddn bind-dn
{ --bindpwd bind-password
| --bindpwdfile filename
}]
{{ --schema filename
| --acl filename
| --udm_syntax filename
| --udm_hook filename
...}
| --udm_module filename
[--messagecatalog filename
...]
[--umcregistration filename
]
[--icon filename
...]
}
[--packagename packagename
] [--packageversion packageversion
] [--ucsversionstart ucsversion
] [--ucsversionend ucsversion
]
The shell function ucs_registerLDAPExtension
from the Univention shell function library (see Section 14.3) can be used to register several extension in LDAP.
This shell function offers several modes:
--schema
filename
.schema
--acl
filename
.acl
--udm_syntax
filename
.py
--udm_hook
filename
.py
--udm_module
filename
.py
The modes can be combined. If more than one mode is used in one call of the function, the modes are always processed in the order as listed above. Each of these options expects a filename as an required argument.
The following options can be given multiple times, but only after the option --udm_module
:
--messagecatalog
prefix
/language
.mo
The option can be used to supply message translation files in GNU message catalog format.
The language must be a valid language tag, i.e. must correspond to a subdirectory of /usr/share/locale/
.
--umcregistration
filename
.xml
The option can be used to supply an UMC registration file (see Section 7.5.2) to make the UDM module accessible via Univention Management Console (UMC).
--icon
filename
The option can be used to supply icon files (png
or jpeg
, in 16×16 or 50×50, or svgz
).
Called from a joinscript, the function automatically determines some required parameters, like the app identifier plus Debian package name and version, required for the creation of the corresponding object. After creation of the object the function waits up to 3 minutes for the master domain controller to signal availability of the new extension and reports success or failure. For UDM extensions it additionally checks that the corresponding file has been made available in the local filesystem. Failure conditions may occur e.g. in case the new LDAP schema extension collides with the schema currently active. The master domain controller only activates a new LDAP schema or ACL extension if the configuration check succeeded.
Before calling the shell function the shell variable UNIVENTION_APP_IDENTIFIER
should be set
to the versioned app identifier (and exported to the environment of subprocesses).
The shell function will then register the specified app identifier with the extension object to indicate that
the extension object is required as long as this app is installed anywhere in the UCS domain.
The options --packagename
and --packageversion
should usually not be used, as these parameters
are determined automatically.
To prevent accidental downgrades the function ucs_registerLDAPExtension
(as well as the corresponding UDM module)
only execute modifications of an existing object if the Debian package version is not older than the previous one.
ucs_registerLDAPExtension
supports two additional options to specify a valid range of UCS versions, where an extension should be
activated. The options are --ucsversionstart
and --ucsversionend
. The version check is only
performed whenever the extension object is modified. By calling this function from a joinscript, it will
automatically update the Debian package version number stored in the object, triggering a re-evaluation of the specified UCS version range.
The extension is activated up to and excluding the UCS version specified by --ucsversionend
.
This validity range is not applied to LDAP schema extensions, since they must not be undefined as long as there are objects
in the LDAP directory which make use of it.
Example 3.4. Extension registration in join script
export UNIVENTION_APP_IDENTIFIER="appID-appVersion" ## example . /usr/share/univention-lib/ldap.sh ucs_registerLDAPExtension "$@" \ --schema /path/to/appschemaextension.schema \ --acl /path/to/appaclextension.acl \ --udm_syntax /path/to/appudmsyntax.py ucs_registerLDAPExtension "$@" \ --udm_module /path/to/appudmmodule.py \ --messagecatalog /path/to/de.mo \ --messagecatalog /path/to/eo.mo \ --umcregistration /path/to/module-object.xml \ --icon /path/to/moduleicon16x16.png \ --icon /path/to/moduleicon50x50.png
ucs_unregisterLDAPExtension
[--binddn bind-dn
{ --bindpwd bind-password
| --bindpwdfile filename
}]
{ --schema objectname
| --acl objectname
| --udm_syntax objectname
| --udm_hook objectname
| --udm_module objectname
...}
There is a corresponding ucs_unregisterLDAPExtension
function, which can be used to unregister extension objects. This only works if no App is registered any longer for the object. It must not be called unless it has been verified that no object in LDAP still requires this schema extension. For this reason it should generally not be called in unjoin scripts.
Example 3.5. Schema unregistration in unjoin script
. /usr/share/univention-lib/ldap.sh ucs_unregisterLDAPExtension "$@" --schema appschemaextension
On package removal packages should clean up the data in Univention directory service.
Removing data from LDAP also requires appropriate credentials, while removing a package only requires local root
privileges.
Therefore UCS provides support for so-called unjoin scripts.
In most cases it reverts the changes of a corresponding join script.
A domain is a distributed system. Just because one local system no longer wants to store some information in Univention directory service does not mean that the data should be deleted. There might still be other systems in the domain which still require the data.
Therefore “the first system to come” should setup the data, while only “the last system to go” may clean up the data.
Just like join scripts an unjoin script is prefixed with a two-digit number for lexicographical ordering.
To reverse the order of the unjoin scripts in comparison to the corresponding join scripts, the number of the unjoin script should be 100 minus the number of the corresponding join script.
The suffix of an unjoin script is -uninstall.uinst
and it should be installed in /usr/lib/univention-uninstall/
.
On package removal the unjoin script would be deleted as well, while the Univention directory service might still contain data managed by the package.
Therefore the script must be copied from there to /usr/lib/univention-install/
in the prerm
maintainer script.
Example:
The package univention-fetchmail provides both a join script /usr/lib/univention-install/91univention-fetchmail.inst
and the corresponding unjoin script as /usr/lib/univention-uninstall/09univention-fetchmail-uninstall.uinst
.
As of UCS 3.1 .inst
and .uinst
are not distinguishable in the UMC Join module by the user.
Therefore it is important to use the -uninstall
suffix to give users a visual hint.
Internally join scripts are always executed before unjoin scripts and then ordered lexicographically by their prefix.
To decide if an unjoin script is the last instance and should remove the data from LDAP, a service can be registered for each host where the package is installed.
For example the package univention-fetchmail uses ucs_addServiceFromLocalhost "Fetchmail" "$@"
in the join script to register and ucs_removeServiceFromLocalhost "Fetchmail" "$@"
in the unjoin script to unregister a service for the host.
The data is removed from LDAP when in the unjoin script ucs_isServiceUnused "Fetchmail" "$@"
returns 0.
As a side effect adding the service also allows using this information to find and list those servers currently providing the Fetchmail service.
50join-template-uninstall.uinst
This unjoin script reverts the changes of the join script from Section 3.4.1.
#!/bin/sh # VERSION is needed for some tools to recognize that as a join script VERSION=1 . /usr/share/univention-join/joinscripthelper.lib joinscript_init SERVICE="MyService" eval "$(ucr shell)" . /usr/share/univention-lib/ldap.sh ucs_removeServiceFromLocalhost "$SERVICE" "$@" || die if ucs_isServiceUnused "$SERVICE" "$@" then # was last server to implement service. now the data # may be removed univention-directory-manager container/cn remove "$@" --dn \ "cn=myservice,cn=custom attributes,cn=univention,$ldap_base" || die # Terminate UDM server to force module reload . /usr/share/univention-lib/base.sh stop_udm_cli_server fi # do NOT call "joinscript_save_current_version" # otherwise an entry will be appended to /var/univenion-join/status # instead the join script needs to be removed from the status file joinscript_remove_script_from_status_file join-template exit 0
debian/join-template.prerm
The unjoin script has to be copied to the join script directory before it gets removed:
#!/bin/sh #DEBHELPER# if [ "$1" = "remove" ] then cp /usr/lib/univention-uninstall/50join-template-uninstall.uinst \ /usr/lib/univention-install/ fi exit 0
debian/join-template.postrm
The unjoin script should be invoked automatically on master and backup domain controller systems after the package is removed.
On all other system roles an administrator must run the join script manually through univention-run-join-scripts
.
#!/bin/sh #DEBHELPER# if [ "$1" = "remove" ] then . /usr/share/univention-lib/all.sh call_unjoinscript 50join-template-uninstall.uinst fi exit 0
debian/join-template.postinst
In case the package is installed again and the unjoin script still exists, because it was never executed, the unjoin script must be removed:
#!/bin/sh #DEBHELPER# if [ "$1" = "configure" ] then uinst=/usr/lib/univention-install/50join-template-uninstall.uinst [ -e "$uinst" ] && rm "$uinst" fi . /usr/share/univention-lib/base.sh call_joinscript 50join-template.inst exit 0
An LDAP server provides authenticated and controlled access to directory objects over the network. LDAP objects consist of a collection of attributes which conform to so called LDAP schemata. An in depth documentation of LDAP is beyond the scope of this document, other sources cover this topic exhaustively, e.g. http://www.zytrax.com/books/ldap/ or the man pages (slapd.conf(5), slapd.access(5)).
At least it should be noted that OpenLDAP offers two fundamentally different tool sets for direct access or modification of LDAP data:
The slap
commands (*
slapcat
, etc.) are very low level, operating directly on the LDAP backend data and
should only be used in rare cases, usually with the LDAP server not running. The ldap
commands (*
ldapsearch
, etc.) on the
other hand are the proper way to perform LDAP operations from the command line and their functionality can equivalently be used from all major
programming languages.
On top of the raw LDAP layer, the Univention Directory Manager offers an object model on a higher level, featuring advanced object semantics (see Chapter 6). That is the level that usually appropriate for app developers, which should be considered before venturing down to the level of direct LDAP operations. On the other hand, for the development of new UDM extensions it is important to understand some of the essential concepts of LDAP as used in UCS.
One essential trait of LDAP as used in UCS, is the strict enforcement of LDAP schemata. An LDAP server refuses to start if an unknown LDAP attribute is referenced either in the configuration or in the backend data. This makes it critically important to install schemata on all systems. To simplify this task UCS features a builtin mechanism for automatic schema replication to all UCS hosted LDAP servers in the UCS domain (see Chapter 5). The schema replication mechanism is triggered by installation of a new schema extension package on the UCS master. For redundancy it is strongly recommended to install schema extension packages also on the UCS backup systems. This way, a UCS backup can replace a UCS master in case the master needs to be replaced for some reason. To simplify these tasks even further, UCS offers methods to register new LDAP schemata and associated LDAP ACLs from any UCS system.
For some purposes, e.g. for app installation, it is convenient to be able to register a new LDAP schema extension
from any system in the UCS domain. For this purpose, the schema extension can be stored as a special type of UDM object. The module
responsible for this type of objects is called settings/ldapschema
. As these objects are replicated throughout the
UCS domain, the master domain controller and backup domain controller systems listen for modifications of these objects and integrate them into the local
LDAP schema directory. As noted above, this simplifies the task of keeping the schema on the backup domain controller systems
up to date with that on the master domain controller.
The commands to create the LDAP schema extension objects in UDM may be put into any join script (see Chapter 3).
A LDAP schema extension object is created by using the UDM command line interface univention-directory-manager
or its alias udm
.
LDAP schema extension objects can be stored anywhere in the LDAP directory, but the recommended location would be cn=ldapschema,cn=univention,
below the LDAP base.
Since the join script creating the attribute may be called on multiple hosts, it is a good idea to add the --ignore_exists
option, which suppresses the error exit code in case the object already exists in LDAP.
The UDM module settings/ldapschema
requires several parameters:
name
(required)data
(required)filename
(required)package
(required)packageversion
(required)appidentifier
(optional)string
.
active
(internal)FALSE
).
Since many of these parameters are determined automatically by the ucs_registerLDAPExtension
shell library function, it is recommended to use the shell library function to create these objects (see Section 3.4.3.2).
Example 4.1. Schema registration in join script
export UNIVENTION_APP_IDENTIFIER="appID-appVersion" ## example . /usr/share/univention-lib/ldap.sh ucs_registerLDAPExtension "$@" \ --schema /path/to/appschemaextension.schema
For some purposes, e.g. for app installation, it is convenient to be able to register a new LDAP ACL extension
from any system in the UCS domain. For this purpose, the UCR template for an ACL extension can be stored as a special type of UDM object.
The module responsible for this type of objects is called settings/ldapacl
. As these objects are replicated throughout the
UCS domain, the master domain controller, backup domain controller and slave domain controller systems listen for modifications on these objects and integrate them into the local
LDAP ACL UCR template directory. This simplifies the task of keeping the LDAP ACLs on the backup domain controller systems
up to date with those on the master domain controller.
The commands to create the LDAP ACL extension objects in UDM may be put into any join script (see Chapter 3).
A LDAP ACL extension object is created by using the UDM command line interface univention-directory-manager
or its alias udm
.
LDAP ACL extension objects can be stored anywhere in the LDAP directory, but the recommended location would be cn=ldapacl,cn=univention,
below the LDAP base.
Since the join script creating the attribute may be called on multiple hosts, it is a good idea to add the --ignore_exists
option, which suppresses the error exit code in case the object already exists in LDAP.
The UDM module settings/ldapacl
requires several parameters:
name
(required)data
(required)filename
(required)package
(required)packageversion
(required)appidentifier
(optional)string
.
ucsversionstart
(optional)ucsversionend
(optional)active
(internal)FALSE
).
Since many of these parameters are determined automatically by the ucs_registerLDAPExtension
shell library function, it is recommended to use the shell library function to create these objects (see Section 3.4.3.2).
Example 4.2. LDAP ACL registration in join script
export UNIVENTION_APP_IDENTIFIER="appID-appVersion" ## example . /usr/share/univention-lib/ldap.sh ucs_registerLDAPExtension "$@" \ --acl /path/to/appaclextension.acl
The credentials for different UCS domain accounts are stored in plain-text files on some UCS systems.
The files are named /etc/
.
They are owned by the user *
.secretroot
and allow read-access for different groups.
/etc/ldap.secret
for cn=admin,$ldap_base
DC Backup Hosts
.
/etc/machine.secret
for $ldap_hostdn
cn=dc,cn=computers,$ldap_base
get additional rights to access LDAP attributes.
The file is available on all joined system roles and is readable only by the local root
user and group.
During package installation, only the maintainer scripts (see Section B.3.5) on master and backup domain controller can use their root
permission to directly read /etc/ldap.secret
.
Thus only on those roles the join scripts get automatically executed when the package is installed.
On all other system roles, the join scripts need to be executed manually.
This can either be done through the UMC Join module or through the command line tool univention-run-join-scripts
.
Both methods require appropriate credentials.
To reconfirm the trust relation between UCS systems, computers need to regularly change the password associated with the machine account.
This is controlled through the Univention Configuration Registry variable server/password/change
.
For UCS servers this is evaluated by the script /usr/lib/univention-server/server_password_change
, which is invoked nightly at 01:00 by cron(8).
The interval is controlled through a second Univention Configuration Registry variable server/password/interval
, which defaults to 21 days.
The password is stored in the plain text file /etc/machine.secret
.
Many long running services read these credentials only on startup, which breaks when the password is changed while they are still running.
Therefore UCS provides a mechanism to invoke arbitrary commands, when the machine password is changed.
This can be used for example to restart specific services.
Hook scripts should be placed in the directory /usr/lib/univention-server/server_password_change.d/
.
The name must consist of only digits, upper and lower ASCII characters, hyphens and underscores.
They file must be executable and is called via run-parts(8).
It receives one argument, which is used to distinguish three phases:
Procedure 4.1. Server password change procedure
Each script will be called with the argument prechange
before the password is changed.
If any script terminates with an exit status unequal zero, the change is aborted.
A new password is generated locally using makepasswd(1).
It is changed in the Univention directory service via UDM and stored in /etc/machine.secret
.
The old password is logged in /etc/machine.secret.old
.
If anything goes wrong in this step, the change is aborted and the changes need to be rolled back.
All hook scripts are called again.
If the password change was successful, postchange
gets passed to the hook scripts.
This should complete any change prepared in the prechange
phase.
If the password change failed for any reason, all hook scripts are called with the argument nochange
.
This should undo any action already done in the prechange
phase.
Example 4.3. Server password change example
Install this file to /usr/lib/univention-server/server_password_change.d/
.
#!/bin/sh case "$1" in prechange) # nothing to do before the password is changed exit 0 ;; nochange) # nothing to do after a failed password change exit 0 ;; postchange) # restart daemon after password was changed invoke-rc.d my-daemon restart ;; esac
init-scripts should only be invoked indirectly through invoke-rc.d(8).
This is required for chroot
environments and allows the policy layer to control starting and stopping in certain special situations like during an system upgrade.
Replication of the directory data within a UCS domain is provided by the Univention Directory Listener/Notifier mechanism:
The active Univention Directory Listener instances in the domain connect to a Univention Directory Notifier service. If an LDAP change is performed on the master domain controller (all other LDAP servers in the domain are read-only), this is registered by the Univention Directory Notifier and reported to the listener instances.
Each Univention Directory Listener instance hosts a range of Univention Directory Listener modules. These modules are shipped by the installed applications; the print server package includes, for example, listener modules which generate the CUPS configuration.
Univention Directory Listener modules can be used to communicate domain changes to services which are not
LDAP-aware. The print server CUPS is an example of this: The printer definitions are not
read from the LDAP, but instead from the file /etc/cups/printers.conf
.
Now, if a printer is saved in the printer management of the Univention Management Console, it is stored
in the LDAP directory. This change is detected by the Univention Directory Listener module
cups-printers and an entry gets added to, modified in or deleted from
/etc/cups/printers.conf
based on the modification in the LDAP directory.
By default the Listener loads all modules from the directory /usr/lib/univention-directory-listener/system/
.
Other directories can be specified using the option -m
when starting the univention-directory-listener
daemon.
Each listener module must declare several string constants. They are required by the Univention Directory Listener to handle each module. They should be defined at the beginning of the module.
name = "module_name" description = "Module description" filter = "(objectClass=*)" attribute = ["objectClass"] modrdn = "1"
name
(required)
This name is used to uniquely identify the module.
This should match with the filename containing this listener module without the .py
suffix.
The name is used to keep track of the module state in /var/lib/univention-directory-listener/handlers/
.
description
(required)A short description. It is currently unused and displayed nowhere.
filter
(required)Defines a LDAP filter which is used to match the objects the listener is interested in. This filter is similar to the LDAP search filter as defined in RFC 2254, but more restricted:
The name filter
has the drawback that it shadows the Python built-in function filter()
, but its use has historical reasons.
If that function is required for implementing the listener module, an alias-reference may be defined before overwriting the name or it may be explicitly accessed via the Python __builtin__
module.
attributes
(optional)A Python list of LDAP attribute names which further narrows down the condition under which the listener module gets called. By default the module is called on all attribute changes of objects matching the filter. If the list is specified, the module is only invoked when at least one of the listed attributes is changed.
modrdn
(optional)
Setting this variable to the string 1
changes the signature of the function handler()
.
It receives an additional 4th argument, which specifies the LDAP operation triggering the change.
In addition to the static strings a module must implement several functions. They are called in different situations of the live-cycle of the module.
def initialize(): pass def handler(dn, new, old[, command='']): pass def clean(): pass def prerun(): pass def postrun(): pass def setdata(key, value): pass
handler(dn, old, new, command='')
(required)
This function is called for each change matching the filter
and attributes
as declared in the header of the module.
The distinguished name (dn) of the object is supplied as the first argument dn
.
Depending on the type of modification, old
and new
may each independently either be None
or reference a Python dictionary of lists.
Each list represents one of the multi-valued attributes of the object.
The Univention Directory Listener uses a local cache to store the values of each object as it has seen most recently.
This cache is used to supply the values for old
, while the values in new
are those retrieved from that LDAP directory service which is running on the same server as the Univention Directory Notifier (master domain controller or backup domain controller servers in the domain).
If and only if the global modrdn
setting is enabled, command
is passed as a fourth argument.
It contains a single letter, which indicates the type of modification.
This can be used to distinguish an modrdn operation from a delete operation followed by a create operation.
m
(modify)
Signals a modify operation, where an existing object is changed.
old
contains a copy of the previously cached values and new
contains the new values as retrieved from the LDAP directory service.
a
(add)
Signals the addition of a new object.
old
is None
and new
contains the values of the added object.
d
(delete)
Signals the removal of a previously existing object.
old
contains a copy of the previously cached values, while new
is None
.
r
(rename: modification of distinguished name via modrdn
)Signals a change in the distinguished name, which may be caused by renaming a object or moving the object from one container into another. The module is called with this command instead of the delete command, so that modules can recognize this special case and avoid deletion of local data associated with the object. The module will be called again with the add command just after the modrdn command, where it should process the rename or move operation. Each module is responsible for keeping track of the rename-case by internally storing the previous distinguished name during the modrdn phase of this two phased operation.
n
(new or schema change)This command can signal two changes:
dn
is cn=Subschema
, it signals that a schema change occurred.add
operation.initialize()
(optional),clean()
(optional)
The function initialize()
is called once when the Univention Directory Listener loads the module for the first time.
This is recorded persistently in the file /var/lib/univention-directory-listener/
, where name
name
equals the value from the header.
If for whatever reason the listener module should be reset and re-run for all matching objects, the state can be reset by running the command univention-directory-listener-ctrl resync
.
In that case the function name
initialize()
will be called again.
The function clean()
is only called when the Univention Directory Listener is initialized for the first time or is forced to “re-initialize from scratch” using the -g
or -i
option.
The function should purge all previously generated files and return the module into a clean state.
prerun()
(optional),postrun()
(optional)
For optimization the Univention Directory Listener does not keep open an LDAP connection all time.
Instead the connection is opened once at the beginning of a change and closed only if no new change arrives within 15 seconds.
The opening is signaled by the invocation of the function prerun()
and the closing by postrun()
.
The function postrun()
is most often used to restart services, as restarting a service takes some time and makes the service unavailable during that time.
It's best practice to use the handler()
only to process the stream of changes, set UCR variables or generate new configuration files.
Restarting associated services should be delayed to the postrun()
function.
The function postrun()
is only called, when no change happens for 15 seconds.
This is not on a per-module basis, but globally.
In an ever changing system, where the stream of changes never pauses for 15 seconds, the functions may never be called!
setdata(key, value)
(optional)
This function is called up to four times by the Univention Directory Listener main process to pass configuration data into the modules.
The following key
s are supplied in the following order:
basedn
binddn
simple bind
).bindpw
ldapserver
It's strongly recommended to avoid initiating LDAP modifications from a listener module. This potentially creates a complex modification dynamic, considering that a module may run on several systems in parallel at their own timing.
All changes trigger a call to the function handle()
.
For simplicity and readability it is advisable to delegate the different change types to different sub-functions.
The following boilerplate code delegates each change type to a separate function. It does not handle renames and moves explicitly, but only as the removal of the object at the old dn and the following addition at the new dn.
Source code: doc/developer-reference/listener/simple.py
def handler(dn, new, old): if new and not old: handler_add(dn, new) elif new and old: handler_modify(dn, old, new) elif not new and old: handler_remove(dn, old) else: pass # ignore def handler_add(dn, new): """Handle addition of object.""" pass # replace this def handler_modify(dn, old, new): """Handle modification of object.""" pass # replace this def handler_remove(dn, old): """Handle removal of object.""" pass # replace this
In case rename and move actions should be handled separately, the following code may be used:
Source code: doc/developer-reference/listener/modrdn.py
modrdn = "1" _delay = None def handler(dn, new, old, command): global _delay if _delay: old_dn, old = _delay _delay = None if "a" == command and old['entryUUID'] == new['entryUUID']: handler_move(old_dn, old, dn, new) return handler_remove(old_dn, old) if "n" == command and "cn=Subschema" == dn: handler_schema(old, new) elif new and not old: handler_add(dn, new) elif new and old: handler_modify(dn, old, new) elif not new and old: if "r" == command: _delay = (dn, old) else: handler_remove(dn, old) else: pass # ignore, reserved for future use def handler_move(old_dn, old, new_dn, dn): """Handle rename or move of object.""" pass # replace this def handler_schema(old, new): """Handle change in LDAP schema.""" pass # replace this
Please be aware that tracking the two subsequent calls for modrdn
in memory might cause duplicates, in case the Univention Directory Listener is terminated while such an operation is performed.
If this is critical, the state should be stored persistently into a temporary file.
The following example shows a listener module, which logs all changes to users into the file /root/UserList.txt
.
Source code: doc/developer-reference/listener/printusers/
""" Example for a listener module, which logs changes to users. """ __package__ = "" # workaround for PEP 366 import listener import os import errno import univention.debug as ud from collections import namedtuple name = 'printusers' description = 'print all names/users/uidNumbers into a file' filter = """\ (& (| (& (objectClass=posixAccount) (objectClass=shadowAccount) ) (objectClass=univentionMail) (objectClass=sambaSamAccount) (objectClass=simpleSecurityObject) (objectClass=inetOrgPerson) ) (!(objectClass=univentionHost)) (!(uidNumber=0)) (!(uid=*$)) )""".translate(None, '\t\n\r') attributes = ['uid', 'uidNumber', 'cn'] _Rec = namedtuple('Rec', ' '.join(attributes)) USER_LIST = '/root/UserList.txt' def handler(dn, new, old): """ Write all changes into a text file. This function is called on each change. """ if new and old: _handle_change(dn, new, old) elif new and not old: _handle_add(dn, new) elif old and not new: _handle_remove(dn, old) def _handle_change(dn, new, old): """ Called when an object is modified. """ o_rec = _rec(old) n_rec = _rec(new) ud.debug(ud.LISTENER, ud.INFO, 'Edited user "%s"' % (o_rec.uid,)) _writeit(o_rec, u'edited. Is now:') _writeit(n_rec, None) def _handle_add(dn, new): """ Called when an object is newly created. """ n_rec = _rec(new) ud.debug(ud.LISTENER, ud.INFO, 'Added user "%s"' % (n_rec.uid,)) _writeit(n_rec, u'added') def _handle_remove(dn, old): """ Called when an previously existing object is removed. """ o_rec = _rec(old) ud.debug(ud.LISTENER, ud.INFO, 'Removed user "%s"' % (o_rec.uid,)) _writeit(o_rec, u'removed') def _rec(data): """ Retrieve symbolic, numeric ID and name from user data. """ return _Rec(*(data.get(attr, (None,))[0] for attr in attributes)) class AsRoot(object): """ Temporarily change effective UID to 'root'. """ def __enter__(self): listener.setuid(0) def __exit__(self, exc_type, exc_value, traceback): listener.unsetuid() def _writeit(rec, comment): """ Append CommonName, symbolic and numeric User-IDentifier, and comment to file. """ nuid = u'*****' if rec.uid in ('root', 'spam') else rec.uidNumber indent = '\t' if comment is None else '' try: with AsRoot(): with open(USER_LIST, 'a') as out: print >> out, u'%sName: "%s"' % (indent, rec.cn) print >> out, u'%sUser: "%s"' % (indent, rec.uid) print >> out, u'%sUID: "%s"' % (indent, nuid) if comment: print >> out, u'%s%s' % (indent, comment,) except IOError as ex: ud.debug( ud.LISTENER, ud.ERROR, 'Failed to write "%s": %s' % (USER_LIST, ex)) def initialize(): """ Remove the log file. This function is called when the module is forcefully reset. """ try: with AsRoot(): os.remove(USER_LIST) ud.debug( ud.LISTENER, ud.INFO, 'Successfully deleted "%s"' % (USER_LIST,)) except OSError as ex: if errno.ENOENT == ex.errno: ud.debug( ud.LISTENER, ud.INFO, 'File "%s" does not exist, will be created' % (USER_LIST,)) else: ud.debug( ud.LISTENER, ud.WARN, 'Failed to delete file "%s": %s' % (USER_LIST, ex))
Some comments on the code:
__package__
is currently necessary, as the Univention Directory Listener imports the listener module by its own mechanism, which is incompatible with the mechanism normally used by Python itself.
Be aware, that this might cause problems when using pickle.
$
-sign.
attribute
filter further restricts the module to only trigger on changes to the numeric and symbolic user identifier and the last name of the user.
tail -f /root/UserList.txt &
.
Then create a new user or modify the lastname of an existing one to trigger the module.
For packaging the following files are required:
debian/printusers.install
The module should be installed into the directory /usr/lib/univention-directory-listener/system/
.
printusers.py usr/lib/univention-directory-listener/system/
debian/printusers.postinst
The Univention Directory Listener must be restarted after package installation and removal:
#! /bin/sh set -e case "$1" in configure) invoke-rc.d univention-directory-listener restart ;; abort-upgrade|abort-remove|abort-deconfigure) ;; *) echo "postinst called with unknown argument \`$1'" >&2 exit 1 ;; esac #DEBHELPER# exit 0
debian/printusers.postrm
#! /bin/sh set -e case "$1" in remove) invoke-rc.d univention-directory-listener restart ;; purge|upgrade|failed-upgrade|abort-install|abort-upgrade|disappear) ;; *) echo "postrm called with unknown argument \`$1'" >&2 exit 1 ;; esac #DEBHELPER# exit 0
For larger modules it might be preferable to use a more object oriented design like the following example, which logs referential integrity violations into a file.
Source code: doc/developer-reference/listener/obj.py
__package__ = "" # workaround for PEP 366 name = "refcheck" description = "Check referential integrity of uniqueMember relations" filter = "(uniqueMember=*)" attribute = ["uniqueMember"] modrdn = "1" import os import ldap import listener import univention.debug as ud from pwd import getpwnam class LocalLdap(object): PORT = 7389 def __init__(self): self.data = {} self.con = None def setdata(self, key, value): self.data[key] = value def prerun(self): try: self.con = ldap.open(self.data["ldapserver"], port=self.PORT) self.con.simple_bind_s(self.data["binddn"], self.data["bindpw"]) except ldap.LDAPError as ex: ud.debug(ud.LISTENER, ud.ERROR, str(ex)) def postrun(self): try: self.con.unbind() self.con = None except ldap.LDAPError as ex: ud.debug(ud.LISTENER, ud.ERROR, str(ex)) class LocalFile(object): USER = "listener" LOG = "/var/log/univention/refcheck.log" def initialize(self): try: ent = getpwnam(self.USER) with AsRoot(): open(self.LOG, "wb") os.chown(self.LOG, ent.pw_uid, -1) except OSError as ex: ud.debug(ud.LISTENER, ud.ERROR, str(ex)) def log(self, msg): with open(self.LOG, 'ab') as log: print >> log, msg def clean(self): try: with AsRoot(): os.remove(self.LOG) except OSError as ex: ud.debug(ud.LISTENER, ud.ERROR, str(ex)) class AsRoot(object): """ Temporarily change effective UID to 'root'. """ def __enter__(self): listener.setuid(0) def __exit__(self, exc_type, exc_value, traceback): listener.unsetuid() class ReferentialIntegrityCheck(LocalLdap, LocalFile): MESSAGES = { (False, False): "Still invalid: ", (False, True): "Now valid: ", (True, False): "Now invalid: ", (True, True): "Still valid: ", } def __init__(self): super(ReferentialIntegrityCheck, self).__init__() self._delay = None def handler(self, dn, new, old, command=''): if self._delay: old_dn, old = self._delay self._delay = None if "a" == command and old['entryUUID'] == new['entryUUID']: self.handler_move(old_dn, old, dn, new) return self.handler_remove(old_dn, old) if "n" == command and "cn=Subschema" == dn: self.handler_schema() elif new and not old: self.handler_add(dn, new) elif new and old: self.handler_modify(dn, old, new) elif not new and old: if "r" == command: self._delay = (dn, old) else: self.handler_remove(dn, old) else: pass # ignore, reserved for future use def handler_add(self, dn, new): if not self._validate(new): self.log("New invalid object: " + dn) def handler_modify(self, dn, old, new): valid = (self._validate(old), self._validate(new)) msg = self.MESSAGES[valid] self.log(msg + dn) def handler_remove(self, dn, old): if not self._validate(old): self.log("Removed invalid: " + dn) def handler_move(self, old_dn, old, new_dn, new): valid = (self._validate(old), self._validate(new)) msg = self.MESSAGES[valid] self.log("%s %s -> %s" % (msg, old_dn, new_dn)) def handler_schema(self): self.log("Schema change") def _validate(self, data): try: for dn in data["uniqueMember"]: self.con.search_ext_s(dn, ldap.SCOPE_BASE, attrlist=[], attrsonly=1) return True except ldap.NO_SUCH_OBJECT: return False except ldap.LDAPError as ex: ud.debug(ud.LISTENER, ud.ERROR, str(ex)) return False _instance = ReferentialIntegrityCheck() initialize = _instance.initialize handler = _instance.handler clean = _instance.clean prerun = _instance.prerun postrun = _instance.postrun setdata = _instance.setdata
The listener runs with the effective permissions of the user listener
.
If root
-privileges are required, listener.setuid()
can be used to switch the effective UID.
When done, listener.unsetuid()
should be called to drop back to the listener
UID.
It's best practice to code this as try
/finally
clauses in Python.
The directory /var/lib/univention-directory-listener/
contains several files:
cache/cache.mdb
, cache/lock.mdb
Starting with UCS 4.2 the LMDB cache database contains a copy of all objects and their attributes.
It is used to supply the old values supplied through the old
parameter, when the function handle()
is called.
The cache is also used to keep track, for which object which module was called. This is required when a new module is added, which is invoked for all already existing objects when the Univention Directory Listener is restarted.
On domain controllers the cache could be replaced by doing a query to the local LDAP server, before the new values are written into it. But member server do not have a local LDAP server, so there the cache is needed. Also note that the cache keeps track of the associated listener modules, which is not available from the LDAP.
cache.lock
Starting with UCS 4.2 this file is used to detect if a listener opened the cache database.
cache.db
, cache.db.lock
Before UCS 4.2 the BDB cache file contained a copy of all objects and their attributes. With the update to UCS 4.2 it gets converted into an LMDB database.
notifier_id
This file contains the last notifier ID read from the Univention Directory Notifier.
handlers/
For each module the directory contains a text file consisting of a single number.
The name of the file is derived from the values of the variable name
as defined in each listener module.
The number is to be interpreted as a bit-field of HANDLER_INITIALIZED=0x1
and HANDLER_READY=0x2
.
If both bits are set, it indicates that the module was successfully initialized by running the function initialize()
.
Otherwise both bits are unset.
The package univention-directory-listener contains several commands useful for controlling and debugging problems with the Univention Directory Listener. This can be useful for debugging listener cache inconsistencies.
univention-directory-listener-ctrl
The command univention-directory-listener-ctrl resync
can be used to reset and re-initialize a module.
It stops any currently running listener process, removes the state file for the specified module and starts the listener process again.
This forces the functions name
clean()
and initialize()
to be called one after the other.
univention-directory-listener-dump
The command univention-directory-listener-dump
can be used to dump the cache file /var/lib/univention-directory-listener/cache.db
.
The Univention Directory Listener must be stopped first by invoking service univention-directory-listener stop
.
It outputs the cache in format compatible to the LDAP Data Interchange Format (LDIF).
univention-directory-listener-verify
The command univention-directory-listener-verify
can be used to compare the content of the cache file /var/lib/univention-directory-listener/cache.db
to the content of an LDAP server.
The Univention Directory Listener must be stopped first by invoking service univention-directory-listener stop
.
LDAP credentials must be supplied at the command line.
For example, the following command would use the machine password:
univention-directory-listener-verify \ -b "$(ucr get ldap/base)" \ -D "$(ucr get ldap/hostdn)" \ -w "$(cat /etc/machine.secret)"
get_notifier_id.py
The command /usr/share/univention-directory-listener/get_notifier_id.py
can be used to get the latest ID from the notifier.
This is done by querying the Univention Directory Notifier running on the LDAP server configured through the Univention Configuration Registry variable ldap/master
.
The returned value should be equal to the value currently stored in the file /var/lib/univention-directory-listener/notifier_id
.
Otherwise the Univention Directory Listener might still be processing a transaction or it might indicate a problem with the Univention Directory Listener
The Listener/Notifier mechanism is used to trigger arbitrary actions when changes occur in the LDAP directory service.
In addition to the LDAP server slapd
it consists of two other services:
The Univention Directory Notifier service runs next to the LDAP server and broadcasts change information to interested parties.
The Univention Directory Listener service listens for those notifications, downloads the changes and runs listener modules performing arbitrary local actions like storing the data in a local LDAP server for replication or generating configuration files for non-LDAP-aware local services.
On startup the listener connects to the notifier and opens a persistent TCP connection. The host can be configured through several Univention Configuration Registry variables:
notifier/server
is explicitly set, only that named host is used.
In addition the Univention Configuration Registry variable notifier/server/port
can be used to explicitly configure a different TCP port other then 6669.
ldap/master
is used.
ldap/master
and ldap/backup
[2].
The following steps occur on changes:
Procedure 5.1. Listener/Notifier procedure
An LDAP object is modified on the master domain controller. Changes initiated on all other system roles are re-directed to the master.
The UCS-specific overlay-module translog
appends the DN to the file /var/lib/univention-ldap/listener/listener
[3].
The Univention Directory Notifier watches that file, picks up and removes each line it processed.
It assigns the next transaction number and writes it into the file /var/lib/univention-ldap/notify/transaction
[4], including the DN and change type.
For efficient access by transaction ID the index transaction.index
is updated.
All listeners get notified of the new transaction.
Each listener triggered in this way queries the Notifier for the latest transaction ID, DN and change type.
Each listener opens a connection to the LDAP server running on the UCS system which was used to query the Notifier. It retrieves the latest state of the object identified through the DN. If access is blocked, for example, by selective replication, the change is handled as a delete operation instead.
The old state of the object is fetched from the local listener cache.
For each module it is checked, if either the old or new state of the object matches the filter
and attributes
specified in the corresponding Python variables.
If not, the module is skipped.
If the function prerun()
of module was not called yet, this is done to signal the start of changes.
The function handler()
specified in the module is called, passing in the DN and the old and new state.
The main listener process updates its cache with the new values, including the names of the modules which successfully handled that object. This guarantees that the module is still called, even when the filter criteria would no longer match the object after modification.
On a backup domain controller the Univention Directory Listener writes the transaction data to the file /var/lib/univention-ldap/listener/listener
[5] to allow the Univention Directory Notifier to be cascaded.
This is configured internally with the option -o
of univention-directory-listener
and is done for load balancing and failover reasons.
The transaction ID is written into the local file /var/lib/univention-directory-listener/notifier_id
.
After 15 seconds of inactivity the function postrun()
is invoked for all prepared modules.
This signals a break in the stream of changes and requests the module to release its resources and/or start pending operations.
[2] This list of backup domain controllers stored in the Univention Configuration Registry variable ldap/backup
is automatically updated by the listener module ldap_server.py
.
[3] Referred to as FILE_NAME_LISTENER
, TRANSACTION_FILE
in the source code
[4] Referred to as FILE_NAME_TF
in the source code
[5] Referred to as FILE_NAME_LISTENER
, TRANSACTION_FILE
in the source code
The Univention Directory Manager (UDM) is a wrapper for LDAP objects. Traditionally LDAP stores objects as a collection of attributes, which are defines by so called schemata. Modifying entries is slightly complicated, as there are no high-level operations to add or remove values from multi-valued attributes, or to keep the password used by different authentication schemes such as Windows NTLM-hashes, UNIX MD5 hashes, or Kerberos tickets in sync.
The command line client udm
provides different modes of operation.
udm
[--binddn bind-dn
--bindpwd bind-password
] [module
] [mode
] [options
]
udm
module
create --set property
=value
...
eval "$(ucr shell)" udm container/ou create --position "$ldap_base" --set name="xxx"
Multiple --set
s may be used to set the values of a multivalued property.
The equivalent LDAP command would look like this:
eval "$(ucr shell)" ldapadd -D "cn=admin,$ldap_base" -y /etc/ldap.secret <<__LDIF__ dn: uid=xxx,$ldap_base objectClass: organizationalRole cn: xxx __LDIF__
udm
module
list [ --dn dn
| --filter property
=value
]
udm container/ou list --filter name="xxx"
univention-ldapsearch cn=xxx
udm
module
modify [ --dn dn
| --filter property
=value
] [ --set property
=value
| --append property
=value
| --remove property
=value
...]
udm container/ou modify --dn "cn=xxx,$ldap_base" --set name="xxx"
For multivalued attributes --append
and --remove
can be used to add additional values or remove existing values.
--set
overwrites any previous value, but can also be used multiple times to specify further values.
--set
and --append
should not be mixed for any property in one invocation.
udm
module
remove [ --dn dn
| --filter property
=value
]
udm container/ou delete --dn "cn=xxx,$ldap_base"
If --filter
is used, it must match exactly one object.
Otherwise udm
refuses to delete any object.
Each UDM module provides a set of mappings from LDAP attributes to properties. This set is not complete, because LDAP objects can be extended with additional auxiliary objectClasses Extended Attributes can be used to extend modules to show additional properties. These properties can be mapped to any already defined LDAP attribute, but objects can also be extended by adding additional auxiliary object classes, which can provide new attributes.
For packing purpose any additional LDAP schema needs to be registered on the master domain controller, which is replicated from there to all other Domaincontrollers via the Listener/Notifier mechanism (see Chapter 5). This is best done trough a separate schema package, which should be installed on the master domain controller and backup domain controller. Since Extended Attributes are declared in LDAP, the commands to create them can be put into any join script (see Chapter 3). To be convenient, the declaration should be also included with the schema package, since installing it there does not require the Administrator to provide additional LDAP credentials.
An Extended Attribute is created by using the UDM command line interface univention-directory-manager
or its alias udm
.
The module is called settings/extended_attribute
.
Extended Attributes can be stored anywhere in the LDAP, but the default location would be cn=custom attributes,cn=univention,
below the LDAP base.
Since the join script creating the attribute may be called on multiple hosts, it is a good idea to add the --ignore_exists
option, which suppresses the error exit code in case the object already exists in LDAP.
The module settings/extended_attribute
requires many parameters.
They are described in ????.
name
(required)CLIName
(required)shortDescription
(required)translationShortDescription
(optional, multiple)longDescription
(required)translationLongDescription
(optional, multiple)objectClass
(required)deleteObjectClass
(optional)ldapMapping
(required)syntax
(optional)string
.
default
(optional)valueRequired
(optional)multivalue
(optional)SINGLE-VALUE
setting of the attribute in the LDAP schema.
mayChange
(optional)notEditable
(optional)hook
(optional)doNotSearch
(optional)tabName
(optional)English
locale.
A new tab is automatically created for new names.
If no name is given, ???
translationTabName
(optional, multiple)tabPosition
(optional)tabName
, for which no tab exists.
The integer value defines the position where the newly tab is inserted.
By default the newly created tab is appended at the end, but before the Extended settings tab.
overwriteTab
(optional)tabName
settings is replaces by a new clean tab with only the properties defined by Extended Attributes.
tabAdvanced
(optional)groupName
(optional)English
locale.
A new tab is automatically created for new names.
If no name is given, the property is placed before the first tab.
translationGroupName
(optional, multiple)groupPosition
(optional)groupName
, for which no group exists.
The integer value defines the position where the newly group is inserted.
By default the newly created group is appended at the end.
overwritePosition
(optional)disableUDMWeb
(optional)fullWidth
(optional)module
(required, multiple)options
(required, multiple)version
(required)2
.
Create the Extended Attribute first through UMC-UDM.
Modify it until you're satisfied.
Only then dump it using udm settings/extended_attribute list
and convert the output to an equivalent shell script creating it.
Example 6.1. Extended Attribute for custom LDAP schema
This example provides a simple LDAP schema called extended-attribute.schema
, which declares one object class univentionExamplesUdmOC
and one attribute univentionExamplesUdmAttribute
.
#objectIdentifier univention 1.3.6.1.4.1.10176 #objectIdentifier univentionCustomers univention:99999 #objectIdentifier univentionExamples univentionCustomers:0 objectIdentifier univentionExamples 1.3.6.1.4.1.10176:99999:0 objectIdentifier univentionExmaplesUdm univentionExamples:1 objectIdentifier univentionExmaplesUdmAttributeType univentionExmaplesUdm:1 objectIdentifier univentionExmaplesUdmObjectClass univentionExmaplesUdm:2 attributetype ( univentionExmaplesUdmAttributeType:1 NAME 'univentionExamplesUdmAttribute' DESC 'An example attribute for UDM' EQUALITY caseIgnoreMatch SUBSTR caseIgnoreSubstringsMatch SYNTAX 1.3.6.1.4.1.1466.115.121.1.15{42} SINGLE-VALUE ) objectClass ( univentionExmaplesUdmObjectClass:1 NAME 'univentionExamplesUdmOC' DESC 'An example object class for UDM' SUP top AUXILIARY MUST ( univentionExamplesUdmAttribute ) )
w The schema is shipped as /usr/share/extended-attribute/extended-attribute.schema
and installed by calling ucs_registerLDAPExtension
from the join-script 50extended-attribute.inst
.
#!/bin/sh #DEBHELPER# . /usr/share/univention-lib/base.sh call_joinscript 50extended-attribute.inst exit 0
This script calls the join-script 50extended-attribute.inst
, which also creates an Extended Attribute by using the udm
command line interface:
#!/bin/bash VERSION=1 . /usr/share/univention-join/joinscripthelper.lib . /usr/share/univention-lib/ldap.sh joinscript_init # register LDAP schema for new extended attribute ucs_registerLDAPExtension "$@" --schema /usr/share/extended-attribute/extended-attribute.schema # Register new service entry for this host eval "$(ucr shell)" udm settings/extended_attribute create "$@" --ignore_exists \ --position "cn=custom attributes,cn=univention,$ldap_base" \ --set name="My Attribute" \ --set CLIName="myAttribute" \ --set shortDescription="Example attribute" \ --set translationShortDescription='"de_DE" "Beispielattribut"' \ --set longDescription="An example attribute" \ --set translationLongDescription='"de_DE" "Ein Beispielattribut"' \ --set tabAdvanced=1 \ --set tabName="Examples" \ --set translationTabName='"de_DE" "Beispiele"' \ --set tabPosition=1 \ --set module="groups/group" \ --set module="computers/memberserver" \ --set syntax=string \ --set default="Lorem ipsum" \ --set multivalue=0 \ --set valueRequired=0 \ --set mayChange=1 \ --set doNotSearch=1 \ --set objectClass=univentionExamplesUdmOC \ --set ldapMapping=univentionExamplesUdmAttribute \ --set deleteObjectClass=0 # --set overwritePosition= # --set overwriteTab= # --set hook= # --set options= # Terminate UDM server to force module reload . /usr/share/univention-lib/base.sh stop_udm_cli_server joinscript_save_current_version exit 0
This example is deliberately missing an unjoin-script (see Section 3.5) to keep this example simple. It should check if the Extended Attribute is no longer used in the domain and then remove it.
Sometimes an Extended Attribute should show a list of options to choose from.
This list can either be static or dynamic.
After defining such a new syntax it can be used by referencing its name in the syntax
property of an Extended Attribute.
The static list of available selections is defined once and can not be modified interactively through UMC.
Such a list is best implemented though a custom syntax class.
As the implementation must be available on all system roles, the new syntax is best registered in LDAP.
This can be done by using ucs_registerLDAPExtension
which is described in Section 3.4.3.2.
As an alternative the file can be put into the directory /usr/share/pyshared/univention/admin/syntax.d/
and linked into the directory /usr/lib/pymodules/python2.6/univention/admin/syntax.d/
.
When included into a Debian package, the linking is normally done by dh_python
.
The following example is comparable to the default example in file /usr/share/pyshared/univention/admin/syntax.d/example.py
:
class StaticSelection(select): choices = [ ('value1', 'Description for selection 1'), ('value2', 'Description for selection 2'), ('value3', 'Description for selection 3'), ]
A dynamic list is implemented as an LDAP search, which is described in Section 6.4.2.
For performance reason it is recommended to implement a class derived from UDM_Attribute
or UDM_Objects
instead of using LDAP_Search
.
The file /usr/share/pyshared/univention/admin/syntax.py
contains several examples.
Example 6.2. Dynamic selection list for Extended Attributes
The idea is to create a container with sub-entries for each selection. This following listing declares a new syntax class for selecting a profession level.
class DynamicSelection(UDM_Objects): udm_modules = ('container/cn',) udm_filter = '(&(objectClass=organizationalRole)(ou:dn:=DynamicSelection))' simple = True # only one value is selected empty_value = True # allow selecting nothing key = '%(name)s' # this is stored label = '%(description)s' # this is displayed regex = None # no validation in frontend error_message = 'Invalid value'
The Python code should be put into a file named DynamicSelection.py
.
The following code registers this new syntax in LDAP and adds some values.
It also creates an Extended Attribute for user objects using this syntax.
syntax='DynamicSelection' base="cn=univention,$(ucr get ldap/base)" udm container/ou create --position "$base" \ --set name="$syntax" --set description='UCS profession level' dn="ou=$syntax,$base" udm container/cn create --position "$dn" \ --set name="value1" --set description='UCS Guru (> 5)' udm container/cn create --position "$dn" \ --set name="value2" --set description='UCS Regular (1..5)' udm container/cn create --position "$dn" \ --set name="value3" --set description='UCS Beginner (< 1)' udm container/cn create --ignore_exists --position "$base" \ --set name='udm_syntax' dn="cn=udm_syntax,$base" udm settings/udm_syntax create --position "$dn" \ --set name="$syntax" --set filename="DynamicSelection.py" \ --set data="$(bzip2 <DynamicSelection.py | base64)" \ --set package="$syntax" --set packageversion="1" udm settings/extended_attribute create --position "cn=custom attributes,$base" \ --set name='Profession' \ --set module='users/user' \ --set tabName='General' \ --set translationTabName='"de_DE" "Allgemein"' \ --set groupName='Personal information' \ --set translationGroupName='"de_DE" "Persönliche Informationen"' \ --set shortDescription='UCS profession level' \ --set translationShortDescription='"de_DE" "UCS Erfahrung"' \ --set longDescription='Select a level of UCS experience' \ --set translationLongDescription='"de_DE" "Wählen Sie den Level der Erfahrung mit UCS"' \ --set objectClass='univentionFreeAttributes' \ --set ldapMapping='univentionFreeAttribute1' \ --set syntax="$syntax" --set mayChange=1 --set valueRequired=0
tabName
and groupName
values must exactly match the values already used in the modules.
If they do not match, a new tab or group is added.
This also applies to the translation:
They must match the already translated strings and must be repeated for every Extended Attribute again and again.
The untranslated strings are best extracted directly from the Python source code of the modules in /usr/share/pyshared/univention/admin/handlers/*/*.py
.
For the translated strings run msgunfmt /usr/share/locale/language-code
/LC_MESSAGES/univention-admin*.mo
.
overwritePosition
values must exactly match the name of an already defined property.
Otherwise UDM will crash.
slapd
will fail to start.
objectClass
es from LDAP objects must be done manually.
Currently UDM does not provide any functionality to remove unneeded object classes or methods to force-remove an object class including all attributes, for which the object class is required.
UDM properties can be enabled and disabled via options. For example all properties of a user related to Samba can be switched on or off to reduce the settings shown to an administrator. If many Extended Attributes are added to a UDM module, it might proof necessary to also create new options. Options are per UDM module.
Similar to Extended Attributes an Extended Option is created by using the UDM command line interface univention-directory-manager
or its alias udm
.
The module is called settings/extended_options
.
Extended Options can be stored anywhere in the LDAP, but the default location would be cn=custom attributes,cn=univention,
below the LDAP base.
Since the join script creating the option may be called on multiple hosts, it is a good idea to add the --ignore_exists
option, which suppresses the error exit code in case the object already exists in LDAP.
The module settings/extended_options
has the following properties:
name
(required)shortDescription
(required)translationShortDescription
(optional, multiple)longDescription
(required)translationLongDescription
(optional, multiple)default
(optional)editable
(optional)module
(required, multiple)objectClass
(optional, multiple)Example 6.3. Extended Option
eval "$(ucr shell)" udm settings/extended_options create "$@" --ignore_exists \ --position "cn=custom attributes,cn=univention,$ldap_base" \ --set name="My Option" \ --set shortDescription="Example option" \ --set translationShortDescription='"de_DE" "Beispieloption"' \ --set longDescription="An example option" \ --set translationLongDescription='"de_DE" "Eine Beispieloption"' \ --set default=0 \ --set editable=0 \ --set module="users/user" \ --set objectClass=univentionExamplesUdmOC
Hooks provide a mechanism to pre- and post-process the values of Extended Attributes.
Normally UDM properties are stored as-is in LDAP attributes.
Hooks can modify the LDAP operations when an object is created, modified, deleted or retrieved.
They are implemented in Python and the file must be placed in the directory /usr/share/pyshared/univention/admin/hooks.d/
[6].
The file name must end with .py
.
The module univention.admin.hook provides the class simpleHook
, which implements all required hook functions.
By default they don't modify any request but do log all calls.
This class should be used as a base class for inheritance.
hook_open(self,
obj);
open
handler just before the current state of all properties is saved.
hook_ldap_pre_create(self,
obj);
list hook_ldap_addlist(self,
obj,
al= []);
(ldap-attribute-name, list-of-values)
which will be used to create the LDAP object.
The method must return the (modified) list.
Notice that hook_ldap_modlist
will also be called next.
hook_ldap_post_create(self,
obj);
hook_ldap_pre_modify(self,
obj);
list hook_ldap_modlist(self,
obj,
ml= []);
(ldap-attribute-name, list-of-new-values)
or three-tuples (ldap-attribute-name, list-of-old-values, list-of-new-values)
.
It will be used to create or modify the LDAP object.
The method must return the (modified) list.
hook_ldap_post_modify(self,
obj);
hook_ldap_pre_remove(self,
obj);
hook_ldap_post_remove(self,
obj);
The following example implements a hook, which removes the object-class univentionFreeAttributes
if the property isSampleUser
is no longer set.
from univention.admin.hook import simpleHook class RemoveObjClassUnused(simpleHook): type = 'RemoveObjClassUnused' def hook_ldap_post_modify(self, obj): """Remove unused objectClass.""" ext_attr_name = 'isSampleUser' class_name = 'univentionFreeAttributes' if obj.oldinfo.get(ext_attr_name) in ('1',) and \ obj.info.get(ext_attr_name) in ('0', None): if class_name in obj.oldattr.get('objectClass', []): obj.lo.modify(obj.dn, [('objectClass', class_name, '')])
After installing the file the hook can be activated by setting the hook
property of an Extended Attribute to RemoveObjClassUnused
:
udm settings/extended_attribute modify \ --dn ... \ --set hook=RemoveObjClassUnused
The development of Univention Directory Manager modules is currently only documented in Univention Wiki (currently only available in German): http://wiki.univention.de/index.php?title=Entwicklung_und_Integration_eigener_Module_in_Univention_Directory_Manager
Every UDM property has a syntax, which is used to check the value for correctness.
Univention Corporate Server already provides several syntax types, which are defined in the Python file /usr/share/pyshared/univention/admin/syntax.py
.
The following list of syntaxes is not complete, for a complete overview the file should be consulted directly.
string
,string64
,OneThirdString
,HalfString
,TwoThirdsString
,FourThirdsString
,OneAndAHalfString
,FiveThirdsString
,TextArea
string_numbers_letters_dots
,string_numbers_letters_dots_spaces
,IA5string
,Upload
,Base64Upload
,jpegPhoto
integer
boolean
,booleanNone
,TrueFalse
,TrueFalseUpper
,TrueFalseUp
yes
and no
or true
and false
.
hostName
,DNS_Name
,windowsHostName
,ipv4Address
,ipAddress
,hostOrIP
,v4netmask
,netmask
,IPv4_AddressRange
,IP_AddressRange
,unixTime
,TimeString
,iso8601Date
,date
GroupDN
,UserDN
,UserID
,HostDN
,DomainController
,Windows_Server
,UCS_Server
,LDAP_Search
,UDM_Objects
,UDM_Attribute
Additional syntax classes can be added by placing a Python file in /usr/share/pyshared/univention/admin/syntax.d/
.
They're automatically imported by UDM.
Sometimes the predefined syntax is inappropriate in some scenarios.
This can be because of performance problems with LDAP searches or the need for more restrictive or lenient value checking.
The latter case might require a change to the LDAP schema, since slapd
also checks the provided values for correctness.
The syntax of UDM properties can be overwritten by using Univention Configuration Registry variables.
For each module and each property the variable directory/manager/web/modules/
can be set to the name of a syntax class.
For example module
/properties/property
/syntaxdirectory/manager/web/modules/users/user/properties/username/syntax
=uid
would restrict the name of users to not contain umlauts.
Since UCR variables only affect the local system, the variables must be set on all systems were UDM is used.
This can be either done through a Univention Configuration Registry policy (see ????) or by setting the variable in the .postinst
script of some package, which is installed on all hosts.
It is often required to present a list of entries to the user, from which she can select one or — in case of a multi-valued property — more entries.
Several syntax classes derived from select
provide a fixed list of choices.
If the set of values is known and fixed, it's best to derive an own class from select
and provide the Python file in /usr/share/pyshared/univention/admin/syntax.d/
.
If on the other hand the list is dynamic and is stored in LDAP, UDM provides three methods to retrieve the values.
UDM_Attribute
This class does a UDM search. For each object found all values of a multi-valued property are returned.
For a derived class the following class variables can be used to customize the search:
dn
skips the search and directly returns the property of the UDM object specified by depends
.
%(property-name
)s
should be used to reference properties.
The special property name $attribute$
is replaced by the value of variable attribute
declared above.
(value, display-string)
, which are added as additional selection options.
True
, the empty value is inserted before all other static and dynamic entries.
attribute
must be set to dn
.
The following example syntax would provide a list of all users with their telephone numbers:
class DelegateTelephonedNumber(UDM_Attribute): udm_module = 'users/user' attribute = 'phone' label_format = '%(displayName)s: %($attribute$)s'
UDM_Objects
This class performs a UDM search returning each object found.
For a derived class the following class variables can be used to customize the search:
dn
, which uses the distinguished name of the object.
None
, which uses the UDM specific description
.
dn
can be used to use the distinguished name.
True
changes the widget to a combo-box widget, which only allows to select a single value.
This should be in-sync with the multivalue
property of UDM properties.
False
disables the Python code and directly uses the attributes returned by the LDAP search.
Several properties can then no longer be used as key or label, as those are not explicitly stored in LDAP but are only calculated by the UDM module.
For example, to get the fully qualified domain name of a host %(name)s.%(domain)s
must be used instead of the calculated property %(fqdn)s
.
UDM_Attribute
.
The following example syntax would provide a list of all servers providing a required service:
class MyServers(UDM_Objects): udm_modules = ( 'computers/domaincontroller_master', 'computers/domaincontroller_backup', 'computers/domaincontroller_slave', 'computers/memberserver', ) label = '%(fqdn)s' udm_filter = 'service=MyService'
LDAP_Search
This is the old implementation, which should only be used, if UDM_Attribute
and UDM_Objects
are not sufficient.
In addition to ease of use it has the drawback that Univention Management Console can not do as much caching, which can lead to severe performance problems.
LDAP search syntaxes can be declared in two equivalent ways:
LDAP_Search
and providing that implementation in /usr/share/pyshared/univention/admin/syntax.d/
.
settings/syntax
.
The Python API uses the following variables:
syntax_name
filter
attribute
shares/share: dn
".
They are used as the human readable label for each element of the choices.
value
shares/share: dn
viewonly
True
the values can not be changed.
addEmptyValue
True
the empty value is add to the list of choices.
appendEmptyValue
addEmptyValue
but added at the end.
Used to automatically choose an existing entry in the frontend.
class MyServers(LDAP_Search): def __init__(self): LDAP_Search.__init__(self, filter=('(&(univentionService=MyService)' '(univentionServerRole=member))'), attribute=( 'computers/memberserver: fqdn', ), value='computers/memberserver: dn' ) self.name = 'LDAP_Search' # required workaround
The UDM API uses the following properties:
name
(required)description
(optional)filter
(required)base
(optional)attribute
(optional, multivalued),ldapattribute
(optional, multivalued)value
(optional),ldapvalue
(optional)viewonly
(optional)1
the values can not be changed.
addEmptyValue
(optional)1
the empty value is add to the list of choices.
eval "$(ucr shell)" udm settings/syntax create "$@" --ignore_exists \ --position "cn=custom attributes,cn=univention,$ldap_base" \ --set name=MyServers \ --set filter='(&(univentionService=MyService) (univentionServerRole=member))' \ --set attribute='computers/memberserver: fqdn' \ --set value='computers/memberserver: dn'
For some purposes, e.g. for app installation, it is convenient to be able to deploy a new UDM hook in the UCS domain
from any system in the domain. For this purpose, a UDM hook can be stored as a special type of UDM object.
The module responsible for this type of objects is called settings/udm_hook
. As these objects are replicated throughout the
UCS domain, the UCS servers listen for modifications on these objects and integrate them into the local UDM.
The commands to create the UDM hook objects in UDM may be put into any join script (see Chapter 3).
Like every UDM object a UDM hook object can be created by using the UDM command line interface univention-directory-manager
or its alias udm
.
UDM hook objects can be stored anywhere in the LDAP directory, but the recommended location would be cn=udm_hook,cn=univention,
below the LDAP base.
Since the join script creating the attribute may be called on multiple hosts, it is a good idea to add the --ignore_exists
option, which suppresses the error exit code in case the object already exists in LDAP.
The module settings/udm_hook
requires several parameters.
Since many of these are determined automatically by the ucs_registerLDAPExtension
shell library function, it is recommended to use the shell library function to create these objects (see Section 3.4.3.2).
name
(required)data
(required)filename
(required)package
(required)packageversion
(required)appidentifier
(optional)string
.
ucsversionstart
(optional)ucsversionend
(optional)active
(internal)FALSE
).
For some purposes, e.g. for app installation, it is convenient to be able to deploy a new UDM module in the UCS domain
from any system in the domain. For this purpose, a UDM module can be stored as a special type of UDM object.
The module responsible for this type of objects is called settings/udm_module
. As these objects are replicated throughout the
UCS domain, the UCS servers listen for modifications on these objects and integrate them into the local UDM.
The commands to create the UDM module objects in UDM may be put into any join script (see Chapter 3).
Like every UDM object a UDM module object can be created by using the UDM command line interface univention-directory-manager
or its alias udm
.
UDM module objects can be stored anywhere in the LDAP directory, but the recommended location would be cn=udm_module,cn=univention,
below the LDAP base.
Since the join script creating the attribute may be called on multiple hosts, it is a good idea to add the --ignore_exists
option, which suppresses the error exit code in case the object already exists in LDAP.
The module settings/udm_module
requires several parameters.
Since many of these are determined automatically by the ucs_registerLDAPExtension
shell library function, it is recommended to use the shell library function to create these objects (see Section 3.4.3.2).
name
(required)newapp/someobject
.
data
(required)filename
(required)newapp/someobject.py
).
messagecatalog
(optional)umcregistration
(optional)icon
(optional)png
, jpeg
or svgz
).
package
(required)packageversion
(required)appidentifier
(optional)string
.
ucsversionstart
(optional)ucsversionend
(optional)active
(internal)FALSE
).
For some purposes, e.g. for app installation, it is convenient to be able to deploy a new UDM syntax in the UCS domain
from any system in the domain. For this purpose, a UDM syntax can be stored as a special type of UDM object.
The module responsible for this type of objects is called settings/udm_syntax
. As these objects are replicated throughout the
UCS domain, the UCS servers listen for modifications on these objects and integrate them into the local UDM.
The commands to create the UDM syntax objects in UDM may be put into any join script (see Chapter 3).
Like every UDM object a UDM syntax object can be created by using the UDM command line interface univention-directory-manager
or its alias udm
.
UDM syntax objects can be stored anywhere in the LDAP directory, but the recommended location would be cn=udm_syntax,cn=univention,
below the LDAP base.
Since the join script creating the attribute may be called on multiple hosts, it is a good idea to add the --ignore_exists
option, which suppresses the error exit code in case the object already exists in LDAP.
The module settings/udm_syntax
requires several parameters.
Since many of these are determined automatically by the ucs_registerLDAPExtension
shell library function, it is recommended to use the shell library function to create these objects (see Section 3.4.3.2).
name
(required)data
(required)filename
(required)package
(required)packageversion
(required)appidentifier
(optional)string
.
ucsversionstart
(optional)ucsversionend
(optional)active
(internal)FALSE
).
[6]
This assumes that the hook file is packaged and linked by dh_pysupport
to /usr/lib/pymodules/python
for Python 2.6 or whatever Python version is used.
If the file is installed manually, it must be placed on a path listed in 2.6
/univention/admin/hooks.d/sys.path
.
The Univention Management Console (UMC) is a service that runs an all UCS systems by default. This service provides access to several system information and implements modules for management tasks. What modules are available on a UCS system depends on the system role and the installed components. Each domain user can log an to the service via a web interface. Depending on the access policies for the user the visible modules for management tasks will differ.
In the following the technical details of the architecture and the Python and JavaScript API for modules are described.
The Univention Management Console service consists of four components. The communication between these components is encrypted using SSL. The architecture and the communication channels is shown in Figure 7.1.
All server-side components of the UMC service are based on the asynchronous framework Python Notifier, that provides techniques for handling quasi parallel tasks based on events. The framework follows three basic concepts:
Further details, examples and a complete API documentation for Python Notifier can be found at the website of Python Notifier.
This protocol is used by the UMC server for external clients and between the UMC server and its UMC module processes.
The protocol is based on a server/client model. The client sends requests to the server that will be answered with a response message by the server.
With a status code in the response message the client can determine the type of result of its request:
Before a client may send request messages to the server that contain commands to execute, the client has to authenticate. After a successful authentication the UMC server determines the permissions for the user defined by policies in the LDAP directory. If the LDAP server is not reachable a local cache is checked for previously discovered permissions. If none of these sources is available the user is prohibited to use any command.
The authentication process within the UMC server uses the PAM service univention-management-console
.
By default, this service uses a cache for credentials if the LDAP server is not available to provide the possibility to access the UMC server also in case of problems with the LDAP server.
The messages, request and response, have the same format that consists of a single header line, one empty line and the body.
The header line contains control information that allows the UMC server to verify the correctness of the message without reading the rest of the message.
The header defines the message type, a unique identifier, the length of the message body in bytes, the command and the mime type of the body.
(REQUEST|RESPONSE)/<id>/<length of body>[/<mime-type>]: <command>[ <arguments>]
By the first keyword the message type is defined.
Supported message types are REQUEST
and RESPONSE
.
Any other type will be ignored.
Separated by a /
the message id follows, that must be unique within a communication channel.
By default it consists of a timestamp and a counter.
The next field is a number, defining the length of the body in bytes, starting to count after the empty line.
Since UMCP 2.0 there is as another field specifying the mime type of the body.
If not given then the guessed value for the mime type is application/json
.
If the body can not be decoded using a JSON parser the message is invalid.
The last two fields define the UMCP command that should be executed by the server. The following commands are supported:
AUTH
GET
SET
COMMAND
REQUEST/123423423-01/42/application/json: COMMAND ucr/query
passes on the module command ucr/query to a UMC module.
The message body may contain one object of any type, e.g. an image, an OpenOffice document or a JSON object. The JSON object is the default type and is the only supported mime type for request messages. It contains a dictionary that has a few predefined keys (for both message types):
options
status
message
flavor
This section contains a few example messages of UMCP 2.0.
Example 7.1. Authentication request
REQUEST/130928961341733-1/147/application/json: AUTH {"username": "root", "password": "univention"}
Example 7.2. Search for users
Request:
REQUEST/130928961341726-0/125/application/json: COMMAND udm/query {"flavor": "users/user", "options": {"objectProperty": "name", "objectPropertyValue": "test1*1", "objectType": "users/user"}}
Response:
RESPONSE/130928961341726-0/1639/application/json: COMMAND udm/query {"status": 200, "message": null, "options": {"objectProperty": "name", "objectPropertyValue": "test1*1", "objectType": "users/user"}, "result": [{"ldap-dn": "uid=test11,cn=users,dc=univention,dc=qa", "path": "univention.qa:/users", "name": "test11", "objectType": "users/user"}, ... {"ldap-dn": "uid=test191,cn=users,dc=univention,dc=qa", "path": "univention.qa:/users", "name": "test191", "objectType": "users/user"}]}
With the new generation of UMC there is also an HTTP server available that can be used to access the UMC server. The web server is implemented as a frontend to the UMC server and translates HTTP POST requests to UMCP commands.
Example 7.3. Authentication request
POST http://10.200.15.31/univention/auth HTTP/1.1 {"options": {"username": "root", "password": "univention"}}
Example 7.4. search for users
Request
POST http://10.200.15.31/univention/command/udm/query HTTP/1.1 {"options": {"container": "all", "objectType":"users/user", "objectProperty":"username", "objectPropertyValue":"test1*1"}, "flavor":"users/user"}
Response
{"status": 200, "message": null, "options": {"objectProperty": "username", "container": "all", "objectPropertyValue": "test1*1", "objectType": "users/user"}, "result": [{"ldap-dn": "uid=test11,cn=users,dc=univention,dc=qa", "path": "univention.qa:/users", "name": "test11", "objectType": "users/user"}, ... {"ldap-dn": "uid=test191,cn=users,dc=univention,dc=qa", "path": "univention.qa:/users", "name": "test191", "objectType": "users/user"}]}
Files for building a UMC module.
debian/package
.umc-modules
dh-umc-module-build
builds translation files.
dh-umc-module-install
installs files.
Configured through debian/
.
package
.umc-modules
Module: module-name Python: umc Definition: umc/module-name.xml Javascript: umc Icons: umc/icons
Module
Internal (?) name of the module.
Python
Directory containing the Python code relative to top-level directory.
Definition
Path to an XML file, which describes the module. See Section 7.5.2 for more information.
Javascript
Directory containing the Java-Script code relative to top-level directory.
Icons
(deprecated)
Directory containing the Icons relative to top-level directory.
Must provide icons in sizes 16×16 (umc/icons/16x16/udm-
) and 50×50 (module
.pngumc/icons/50x50/udm-
) pixels.
module
.png
umc/
module
.xml
<?xml version="1.0" encoding="UTF-8"?> <!--DOCTYPE umc SYSTEM "branches/ucs-4.1/ucs-4.1-0/management/univention-management-console/doc/module.dtd"--> <umc version="2.0"> <module id="udm" icon="udm-MODULE" version="1.0" translationId="MODULE"> <name>...</name> <description>...</description> <flavor>...</flavor> <categories> <category name="domain"/> </categories> <command>...</command> </module> </umc>
umc/categories/
category
.xml
<?xml version="1.0" encoding="UTF-8"?> <umc version="2.0"> <categories> <category id="category" priority="..." icon="....svg" color="#xxxxxx"/> </categories> </umc>
The UMC server provides management services that are provided by so called UMC modules. These modules are implemented in Python (backend) and in JavaScript (web frontend). The following page provides information about developing and packaging of UMC modules. It is important to know the details of Section 7.1.
The package univention-management-console-dev provides the command umc-create-module
, which can be used to create a template for a custom UMC module.
The Python API for the UMCP is defined in the python module univention.management.console.protocol
.
A UMC module consists of three components
The UMC server knows three types of resources that define the functionality it can provide:
provide commands that can be executed if the required permission is given.
can be used to verify the correctness of command attributes defined by the UMCP client in the request message or return values provided by the UMC modules.
help to define a structure and to sort the UMC modules by its type of functionality.
All these resources are defined in XML files. The details are described in the following sections
The UMC server does not load the Python modules to get the details about the modules name, description and functionality.
Therefor each UMC module must provide an XML file containing this kind of information.
The following example defines a module with the id udm
:
<?xml version="1.0" encoding="UTF-8"?> <umc version="2.0"> <module id="udm" icon="udm/module" version="1.0"> <name>Univention Directory Manager</name> <description>Manages all UDM modules</description> <flavor icon="udm-users" id="users/user"> <name>Users</name> <description>Managing users</description> </flavor> <categories> <category name="domain"/> </categories> <command name="udm/query" function="query"/> <command name="udm/containers" function="containers"/> </module> </umc>
The element module
defines the basic details of a UMC module.
This identifier must be unique among the modules of an UMC server. Other files may extend the definition of a module by adding more flavors or categories.
The value of this attribute defines an identifier for the icon that should be used for the module. Details for installing icons can be found in the Section 7.6.3.
The child elements name
and description
define the English human readable name and description of the module.
For other translations the build tools will create translation files.
Details can be found in the Section 7.6.3.
This example defines a so called flavor. A flavor defines a new name, description and icon for the same UMC module. This can be used to show several virtual modules in the overview of the web frontend. Additionally the flavor is passed to the UMC server with each request i.e. the UMC module has the possibility to act differently for a specific flavor.
As the next element categories
is defined in the example.
The child elements category
set the categories within the overview where the module should be shown.
Each module can be part of multiple categories.
The attribute name
is the internal identify of a category.
At the end of the definition file a list of commands is specified. The UMC server only passes commands to a UMC module that are defined. A command definition has two attributes:
is the name of the command that is passed to the UMC module.
Within the UMCP message it is the first argument after the UMCP COMMAND
.
defines the method to be invoked within the python module when the command is called.
The predefined set of categories can be extended by each module.
Example 7.5. UMC module category examples
<?xml version="1.0" encoding="UTF-8"?> <umc version="2.0"> <categories> <category id="favorites"> <name>Favorites</name> </category> <category id="system"> <name>System</name> </category> <category id="wizards"> <name>Wizards</name> </category> <category id="monitor"> <name>Surveillance</name> </category> </categories> </umc>
The Python API for UMC modules primary consists of one base class that must be implemented. As an addition to python API provides some helper:
In the definition file for the UMC module specifies functions for the commands provided by the module.
These functions must be implemented as methods of the class Instance
that inherits from univention.management.console.base.Base
.
The following Python code example matches the definition in the previous section:
from univention.management.console import Translation from univention.management.console.base import Base, UMC_Error from univention.management.console.log import MODULE from univention.management.console.config import ucr from univention.management.console.modules.sanitizers import IntegerSanitizer from univention.management.console.modules.decorators import sanitize _ = Translation('univention-management-console-modules-udm').translate class Instance(Base): def init(self): """Initialize the module with some values""" super(Instance, self).init() self.data = [int(x) for x in ucr.get('some/examle/ucr/variable', '1,2,3').split(',')] def query(self, request): """get all values of self.data""" self.finished(request.id, self.data) @sanitize(item=IntegerSanitizer(required=True)) def get(self, request): """get a specific item of self.data""" try: item = self.data[request.options['item']] except IndexError: MODULE.error('A invalid item was accessed.') raise UMC_Error(_('The item %d does not exists.') % (request.options['item'],), status=400) self.finished(request.id, self.data[item]) @sanitize(IntegerSanitizer(required=True)) def put(self, request): """replace all data with the list provided in request.options""" self.data = request.options self.finished(request.id, None)
Each command methods has one parameter that contains the UMCP request. Such an object has the following properties:
id
options
flavor
None
.
The method init()
in the example is invoked when the module process starts. It could e.g. be used to initialize a database connection.
The other methods in the example will serve specific request. To respond to a request the function finished
must be invoked.
To validate the request body the decorator @sanitize
might be used with various sanitizers defined in univention.management.console.modules.sanitizers
.
For a way to send an error message back to the client the UMC_Error
can be raised with the error message as argument and an optional HTTP status code.
The base class for modules provides some properties and methods that could be useful when writing UMC modules:
username
user_dn
password
init()
destroy()
In order to encapsulate and ease the access to module data from the JavaScript side, a store object offers a unified way to query and modify module data. The UMC JavaScript API comes with an object store implementation of the Dojo store API. This allows the JavaScript code to easily access/modify module data and to observe changes on the data in order to react immediately. The following methods are supported:
get(id)
put(dictionary, options)
add(dictionary, options)
remove(id)
query(dictionary)
The UMC object store class in JavaScript will be able to communicate directly with the Python module if the following methods are implemented:
module
/get
module
/put
module
/add
module
/put
.
module
/remove
module
/query
Further references:
A UMC module consists of several files that must be installed at a specific location. As this mechanism is always the same there are debhelper tools making package creation for UMC modules very easy.
The following example is based on the package for the UMC module UCR.
A UMC module may be part of a source package with multiple binary packages. The examples uses a own source package for the module.
As a first step create a source package with the following directories and files:
univention-management-console-module-ucr/
univention-management-console-module-ucr/debian/
univention-management-console-module-ucr/debian/univention-management-console-module-ucr.umc-modules
univention-management-console-module-ucr/debian/rules
univention-management-console-module-ucr/debian/changelog
univention-management-console-module-ucr/debian/control
univention-management-console-module-ucr/debian/copyright
univention-management-console-module-ucr/debian/compat
All these files are standard Debian packaging files except univention-management-console-module-ucr.umc-modules
.
This file contains information about the locations of the UMC module source files:
Module: ucr Python: umc/python Definition: umc/ucr.xml Syntax: umc/syntax/ucr.xml Javascript: umc/js Icons: umc/icons
The keys in this file of the following meaning:
Module
Python
Definition
Javascript
Icons
Syntax
(optional)Category
(optional)The directory structure for such a UMC module file would look like this:
univention-management-console-module-ucr/umc/
univention-management-console-module-ucr/umc/syntax/
univention-management-console-module-ucr/umc/syntax/ucr.xml
univention-management-console-module-ucr/umc/js/
univention-management-console-module-ucr/umc/js/ucr.js
univention-management-console-module-ucr/umc/js/de.po
univention-management-console-module-ucr/umc/de.po
univention-management-console-module-ucr/umc/icons/
univention-management-console-module-ucr/umc/icons/16x16/
univention-management-console-module-ucr/umc/icons/16x16/ucr.png
univention-management-console-module-ucr/umc/icons/24x24/
univention-management-console-module-ucr/umc/icons/24x24/ucr.png
univention-management-console-module-ucr/umc/icons/64x64/
univention-management-console-module-ucr/umc/icons/64x64/ucr.png
univention-management-console-module-ucr/umc/icons/32x32/
univention-management-console-module-ucr/umc/icons/32x32/ucr.png
univention-management-console-module-ucr/umc/ucr.xml
univention-management-console-module-ucr/umc/python/
univention-management-console-module-ucr/umc/python/ucr/
univention-management-console-module-ucr/umc/python/ucr/de.po
univention-management-console-module-ucr/umc/python/ucr/__init__.py
If such a package has been created a few things need to be adjusted
debian/compat
7
debian/rules
%: dh $@ override_dh_auto_build: dh-umc-module-build dh_auto_build override_dh_auto_install: dh-umc-module-install dh_auto_install
debian/control
Source: univention-management-console-module-ucr Section: univention Priority: optional Maintainer: Univention GmbH <packages@univention.de> Build-Depends: debhelper (>= 7.0.50~), python-support, univention-management-console-dev, python-all Standards-Version: 3.5.2 XS-Python-Version: all Package: univention-management-console-module-ucr Architecture: all Depends: univention-management-console-server Description: UMC module for UCR This package contains the UMC module for Univention Configuration Registry
Done through flavor.
<?xml version="1.0" encoding="UTF-8"?> <umc version="2.0"> <module id="udm" icon="udm-MODULE" version="1.0" translationId="MODULE"> <flavor priority="25" icon="udm-MODULE-SUBMODULE" id="MODULE/SUBMODULE"> <name>MODULE name</name> <description>MODULE description</description> </flavor> <categories> <category name="domain"/> </categories> </module> </umc>
Must use /umc/module/category/@name="domain"
!
Must use /umc/module/@translationId
to specify alternative translation file, which must be installed as /usr/share/univention-management-console/i18n/
.
language
/module
.mo
To disabling a module use the following XML file as a template:
<?xml version="1.0" encoding="UTF-8"?> <umc version="2.0"> <module id="udm" icon="udm/module" version="1.0" translationId="MODULE"> <name/> <description/> <flavor id="MODULE/SUBMODULE" deactivated="yes" /> </module> </umc>
When a user opens http://localhost/
or
http://
in a browser, she is redirected to the
UCS overview page.
hostname
/
Depending on the preferred language negotiated by the web browser the user is either redirected to the German or English version. The overview page is split between
and entries.The start page can be extended using Univention Configuration Registry variables. PACKAGE refers to a unique identifier, typically the name of the package shipping the extensions to the overview page. The configurable options are explained below:
ucs/web/overview/entries/admin/PACKAGE/OPTION
variables extend the
administrative section.
ucs/web/overview/entries/service/PACKAGE/OPTION
variables extend the
web services section.
To configure an extension of the overview page the following options must/can be set using the
pattern
ucs/web/overview/entries/admin/PACKAGE/OPTION
=VALUE (and
likewise for services).
link
defines a link to a URL representing the service (usually a web interface).
label
specifies a title for an overview entry. The title
can also be translated; e.g. label/de
can be used for a title
in German.
description
configures a longer description of an overview entry. The description
can also be translated; e.g. description/de
can be used for a description
in German. Should not exceed 60 characters, because of space limitations of the rendered box.
icon
either a filename or a URI
can be provided. When specifying a filename, the name must be relative to the directory
/var/www
, i.e. with a leading '/'. All file formats typically
displayed by browsers can be used (e.g. PNG/JPG). All icons must be scaled to 50x50 pixels.
priority
. Depending on the values
the entries are displayed in lexicographical order (i.e.
100
< 50
).
The following example configures the link to the Nagios web interface:
ucs/web/overview/entries/admin/nagios/description/de: Netzwerk-, Host- und Serviceüberwachung ucs/web/overview/entries/admin/nagios/description: Network, host and service monitoring system ucs/web/overview/entries/admin/nagios/icon: /icon/50x50/nagios.png ucs/web/overview/entries/admin/nagios/label/de: Univention Nagios ucs/web/overview/entries/admin/nagios/label: Univention Nagios ucs/web/overview/entries/admin/nagios/link: /nagios/ ucs/web/overview/entries/admin/nagios/priority: 50
The Univention App Center provides a platform for software vendors and an easy-to-use entry point for Univention Corporate Server users to extend their environment with business software.
The documentation how to develop Apps for Univention App Center can be found in the App Center Developer Guide
Sometimes it might be necessary to add external repositories, e.g. when testing an application which is developed for the UCS@school. Such components can be registered via Univention Management Console or in Univention Configuration Registry.
Components can be versioned. This ensures that only components are installed that are compatible with a UCS version.
current
A list of the integrated repository components is in the UMC module
. Applications which have been added via the Univention App Center are still listed here, but should be managed via the module.A further component can be set up with
. The identifies the component on the repository server. A free text can be entered under , for example, for describing the functions of the component in more detail.The host name of the download server is to be entered in the input field
, and, if necessary, an additional file path in .A
and can be configured for repository servers which require authentication.A software component is only available once
has been activated.A differentiation is also made for components between maintained and unmaintained components.
The following Univention Configuration Registry variables can be used to register a repository component. It is also possible to activate further functions here which cannot be configured via the UMC module. NAME stands for the component's name:
repository/online/component/NAME/server
repository/online/server
uses.
repository/online/component/NAME
repository/online/component/NAME/localmirror
repository/online/component/NAME/server
, a configuration
can be set up so that the component is mirrored, but not activated, or that it
is activated, but not mirrored.
repository/online/component/NAME/description
repository/online/component/NAME/prefix
repository/online/component/NAME/username
repository/online/component/NAME/password
repository/online/component/NAME/version
repository/online/component/NAME/defaultpackages
By default UCS includes English and German localizations. Univention provides a set of tools that facilitates the process of creating translations for Univention Management Console.
This section describes all steps necessary to create a working translation package for UCS. We recommend having a running UCS installation where the tools can be set up in an easy manner. Further more a current GIT checkout of the UCS source code is required.
The package univention-ucs-translation-template contains all tools required to setup and update a translation package. It requires some additional Debian tools to build the package. Run the following command on your UCS to install all needed packages.
sudo univention-install univention-ucs-translation-template dpkg-dev git
The GIT repository is later processed to get initial files for a new translation(often referred to as PO file or Portable Objects).
mkdir ~/translation cd ~/translation git clone \ --single-branch --depth 1 --shallow-submodules \ https://github.com/univention/univention-corporate-server
To create a new translation package for, e.g., French in the current working directory, the following command must be executed:
cd ~/translation univention-ucs-translation-build-package \ --source ~/translation/univention-corporate-server \ --languagecode fr \ --locale fr_FR.UTF-8:UTF-8 \ --language-name French
This creates a new directory ~/translation/univention-l10n-fr/
which contains a Debian source package of the same name.
It includes all source and target files for the translation.
The translation source files (.po
files) are located below the directory ~/translation/univention-l10n-fr/fr
.
Each file should be edited to create the translation.
These files are generated by the package gettext. The manual can be found at http://www.gnu.org/software/gettext/manual/gettext.html. Translation files created by gettext consist of a header and various entries of the form
#: umc/app.js:637 #, python-format msgid "The %s will expire in %d days and should be renewed!" msgstr ""
The first line provides a hint, were the text is used.
The second line is optional and contains flags, which indicate the type and state of the translation.
The string fuzzy
indicates an entry, which was copied by gettext from a previous version and needs to be updated.
The line starting with msgid
contains the original text.
The translation has to be placed on the line containing msgstr
.
Long texts can be split over multiple lines, were each line must start and end with a double-quote.
The following example from the German translation shows a text spanning two lines, with the placeholder present in the original and translated text.
#: umc/js/appcenter/AppCenterPage.js:1067 #, python-format msgid "" "If everything else went correct and this is just a temporary network " "problem, you should execute %s as root on that backup system." msgstr "" "Wenn keine weiteren Fehler auftraten und dies nur ein temporäres " "Netzwerkproblem ist, sollten Sie %s als root auf dem Backup System ausführen."
Some lines contain parameters, in this example %s
and %d
.
They are indicated by a flag like c-format
or python-format
, which must not be removed.
The placeholders have to be carried over to the translated string unmodified and in the same order.
Some other files contain placeholders of the form %{text}s
, which are more flexible and can be reordered.
After a file has been translated completely, the line containing fuzzy
at the beginning of the entry should be removed to avoid warnings.
If a translation string consists of multiple lines the translated string should roughly contain as many lines as the original string.
First update your GIT checkout:
cd ~/translation/univention-corporate-server git pull --rebase
If changes affecting translations are made in the GIT repository, existing translation packages need to be updated to reflect those changes.
Given a path to an updated GIT checkout, univention-ucs-translation-merge
can update a previously created translation source package.
The following example will update the translation package univention-l10n-fr/:
univention-ucs-translation-merge \ ~/translation/univention-corporate-server \ ~/translation/univention-l10n-fr
Before using the new translation, the Debian package has to be built and installed. This can be done with the following commands:
cd ~/translation/univention-l10n-fr sudo apt-get build-dep . dpkg-buildpackage -uc -us -b -rfakeroot sudo dpkg -i ../univention-l10n-fr_*.deb
After logging out of the Univention Management Console the new language should now be selectable in the Univention Management Console login window. Untranslated strings will be still shown in their original language, i.e. in English.
The Univention Updater is used for updating the software. It is based on the Debian APT tools. On top of that the updater provides some UCS specific additions.
UCS releases are provided either via DVD images or via online repositories.
For each major, minor and patchlevel release there is a separate online repository.
They are automatically added to the files in /etc/apt/sources.list.d/
depending on the Univention Configuration Registry variables version/version
and version/patchlevel
, which are managed by the updater.
Separate repositories are used to prevent automatic updates of software packages. This is done to encouraged users to thoroughly test a new release before their systems are updated. The only exception from this rule are the errata updates, which are put into a single repository, which is updated incrementally.
Therefore the updater will include the repositories of a new release in a file called /etc/apt/sources.list.d/00_ucs_temporary_errata__components_update.list
and then do the updates.
Only at the end of a successful update are the Univention Configuration Registry variables updated.
Additional components can be added as separate repositories using Univention Configuration Registry variables repository/online/component/…
, which are described in ???? manual.
Setting the variable …/version=current
can be used to mark a component as required, which blocks an upgrade until the component is available for the new release.
In addition to the regular Debian Maintainer Scripts (see Section B.3.5) the UCS updater supports additional scripts, which are called before and after each release update. Each UCS release and each component can include its own set of scripts.
preup.sh
These scripts is called before the update is started. If any of the scripts aborts with an exit value unequal zero, the update is canceled and never started. The scripts receives the version number of the next release as an command line argument.
For components their preup.sh
scripts is called twice:
Once before the main release preup.sh
script is called and once more after the main script was called.
This is indicated by the additional command line argument pre
respectively post
, which is inserted before the version string.
postup.sh
These scripts is called after the update successfully completed. If any of the scripts aborts with an exit value unequal zero, the update is canceled and does not finish successfully. The scripts receives the same arguments as described above.
The scripts are located in the all/
component of each release and component.
For UCS-4.2 this would be 4.2/maintained/4.2-0/all/preup.sh
and 4.1/maintained/components/
for the some-component
/all/preup.shpreup.sh
script.
The same applies to the postup.sh
script.
The full process is shown in Procedure 12.1.
From UCS 3.2 on the scripts must be digitally signed by an PGP (Pretty Good Privacy) key stored in the key-ring of apt-key(8).
The detached signature must be placed in a separate file next to each updater scripts with the additional file name extension .gpg
, that is preup.sh.gpg
and postup.sh.gpg
.
These extra files are downloaded as well and any error in doing so and in the validation process aborts the updater immediately.
The signatures must be updated after each change to the underlying scripts.
This can be automated or be done manually with a command like the following:
gpg -a -u
key-id
--passphrase-file key-phrase-file
-o script
.sh.gpg -b script
.sh
Signatures can be checked manually using the following command:
gpgv --keyring /etc/apt/trusted.gpg
script
.sh.gpg script
.sh
For an release update the following steps are performed. It assumes a single component is enabled. If multiple components are enabled, the order in which their scripts are called is unspecified. It shows which scripts are called in which order with which arguments.
Procedure 12.1. Update process steps
Create temporary source list file 00_ucs_temporary_errata__components_update.list
Download the preup.sh
and postup.sh
files for the next release and all components into a temporary directory and validate their PGP signatures
Execute component-preup.sh
pre
$version
Execute release-preup.sh
$version
Execute component-preup.sh
post
$version
Download the new Packages
and Release
files. Their PGP signatures validated by APT internally.
Preform the update
Execute component-postup.sh
pre
$version
Execute release-postup.sh
$version
Execute component-postup.sh
post
$version
Set the release related Univention Configuration Registry variables to the new version
UCS provides Single Sign-On functionality with a SAML 2.0 compatible identity provider based on simplesamlphp.
The identity provider is by default installed on the DC Master and all DC Backup servers.
A DNS Record for all systems providing Single Sign-On services is registered for failover, usually ucs-sso.domainname
.
Clients are required to be able to resolve the Single Sign-On DNS name.
New service providers can be registered by using the UDM module saml/serviceprovider
.
To create a new service provider entry in a joinscript, see the following example:
eval "$(ucr shell)" udm saml/serviceprovider create "$@" \ --ignore_exists \ --position "cn=saml-serviceprovider,cn=univention,$ldap_base" \ --set isActivated=TRUE \ --set Identifier="MyServiceProviderIdentifier" \ --set NameIDFormat="urn:oasis:names:tc:SAML:1.1:nameid-format:unspecified" \ --set simplesamlAttributes="false" \ --set AssertionConsumerService="https://$hostname.$domainname/sso-login-page" \ --set simplesamlNameIDAttribute="uid" \ --set privacypolicyURL="https://example.com/policy.html" \ --set serviceProviderOrganizationName="My Service Name" \ --set serviceproviderdescription="A long description shown to the user on the Single Sign-On page." || die
The service provider usually requires at least a public certificate or XML metadata about the identity provider. The certificate can e.g. be downloaded with the following call:
eval "$(ucr shell)" wget --ca-certificate /etc/univention/ssl/ucsCA/CAcert.pem \ https://"${ucs_server_sso_fqdn:-ucs-sso.$domainname}"/simplesamlphp/saml2/idp/certificate \ -O /etc/idp.cert
The XML metadata is available e.g. from
eval $(ucr shell) wget --ca-certificate /etc/univention/ssl/ucsCA/CAcert.pem \ https://"${ucs_server_sso_fqdn:-ucs-sso.$domainname}"/simplesamlphp/saml2/idp/metadata.php \ -O /etc/idp.metadata
The Single Sign-On Login page to be configured in the service provider is https://ucs-sso.domainname/simplesamlphp/saml2/idp/SSOService.php
To provide users with a convenient link to an identity provider initiated login, the following ucr command may be used
fqdn="ucs-sso.domainname" myspi="MyServiceProviderIdentifier" ucr set ucs/web/overview/entries/service/SP/description="External Service Login" \ ucs/web/overview/entries/service/SP/label="External Service SSO" \ ucs/web/overview/entries/service/SP/link="https://$fqdn/simplesamlphp/saml2/idp/SSOService.php?spentityid=$myspi" \ ucs/web/overview/entries/service/SP/description/de="Externer Dienst Login" \ ucs/web/overview/entries/service/SP/label/de="Externer Dienst SSO" \ ucs/web/overview/entries/service/SP/priority=50
Where MyServiceProviderIdentifier is the identifier used when creating the UDM service provider object.
UCS ships with two major database management systems, which are used for UCS internal purposes, but can also be used for custom additions.
UCS uses PostgreSQL by default for its package tracking database, which collects the state and versions of packages installed on all systems of the domain.
By default the MySQL root password is set to ___. Debian provides the dbconfig package, which can be used to create and modify additional databases from maintainer scripts.
Use ucslint
to find packaging mistakes.
Called best from debian/rules
, needs build dependency on ucslint.
override_dh_auto_test: dh_auto_test ucslint
For each issue, ucslint
prints one line, which line contains several fields separated by :
:
severity
:module-id
-test-id
[:filename
[:line-number
[:column-number
]]]:message
For some issues extra context data is printed on the following lines, which are indented with space characters. All other lines start with a letter specifying the severity:
E
W
I
S
The severities are ordered by importance.
By default ucslint
only aborts on errors, but this can be overwritten using the --exitcode-categories
argument followed by a subset of the characters EWIS
.
After the severity an identifier follows, which uniquely identifies the module and the test. The module is given as four digits, which is followed by a dash and the number of the test in that module. Currently the following modules exist:
debian/changelog
file for conformance with Univention rulesdebian/control
file for errors/bin/sh
for BASH constructsThe module and test number may be optionally followed by a file name, line number in that file, and column number in that line, where the issue was found. After that a message is printed, which describes the issue in more detail.
Since ucslint
is very Univention centric, many of its tests return false positives for software packages by other parties.
Therefore many tests need to be disables.
For that a file debian/ucslint.overrides
can be created with list of modules and test, which should be ignored.
Without specifying the optional filename, line number and column number, the test is globally disabled for all files.
The source package univention-lib provides two binary packages shell-univention-lib and python-univention-lib, which contain common library functions usable in shell or Python programs.
This package provides several libraries in /usr/share/univention-lib/
, which can be used in shell scripts.
/usr/share/univention-lib/admember.sh
This file contains some helpers to test for and to manage hosts in AD member mode.
/usr/share/univention-lib/base.sh
This file contains some helpers to create log files, handle unjoin scripts (see Section 3.5) or query the network configuration.
/usr/share/univention-lib/ldap.sh
This file contains some helpers to query data from LDAP, register and unregister service entries, LDAP schema and LDAP ACL extensions.
/usr/share/univention-lib/samba.sh
This file contains a helper to check is Samba4 is used.
/usr/share/univention-lib/ucr.sh
This file contains some helpers to handle bool'ean Univention Configuration Registry variables and handle UCR files on package removal.
/usr/share/univention-lib/umc.sh
This file contains some helpers to handle UMC (see Chapter 7) related tasks.
/usr/share/univention-lib/all.sh
This is a convenient library, which just includes all libraries mentioned above.
This package provides several Python libraries located in the module univention.lib.
This module contains functions to test for and to manage hosts in AD member mode.
This module contains functions to handle at-jobs.
This module provides the function formatTraceback
, which returns the full stack trace for an exception.
This module provides some functions for handling the file /etc/fstab
.
/usr/share/pyshared/univention/lib/getMailFromMailOrUid.py
{ uid | email }
This program returns the distinguished name of the user, which either matches the user identifier or email address given to the command as an argument.
This module provides some classes to handle texts and their translations.
This module provides some helper functions internally used to register LDAP extension as described in Section 3.4.3.2.
This module provides some helper functions internally used by the Directory Listener module handling file shares.
This module provides some functions to implement mutual exclusion using file objects as locking objects.
This module provides miscellaneous functions to query the set of configured LDAP servers, localized domain user names, and other functions.
This module provides some wrappers for dpkg
and APT, which add functions for progress reporting.
This module provides some well known SIDs and RIDs.
This module provides two functions for escaping shell command line arguments and creating at jobs.
This module provides some helper functions internally used for parsing the Univention Configuration Registry variables related to logrotate(8).
This module provides the class UCS_Version
to more easily handle UCS version strings.
This module provides the class Client
to handle connections to remote UMC servers.
This module provides some functions for handling icons.
This module provides a pack-port of urllib2 from Python-3.3, which implements proper certificate checking.
Access control to services can be configured for individual services by setting certain Univention Configuration Registry variables.
Setting auth/
to SERVICE
/restricttrue
enables access control for that service.
This will include the file /etc/security/access-
, which contains the list of allowed users and groups permitted to login to the service.
Users and groups can be added to that file by setting SERVICE
.confauth/
and SERVICE
/user/USER
auth/
to SERVICE
/group/GROUP
true
respectively.
Firewall rules are setup by univention-firewall and can be configured through Univention Configuration Registry or by providing additional UCR templates.
Besides predefined service definitions, Univention Firewall also allows the implementation of package filter rules via Univention Configuration Registry.
These rules are included in /etc/security/packetfilter.d/
via a Univention Configuration Registry module.
Filter rules can be provided via packages or can be configured locally by the administrator. Local rules have a higher priority and overwrite rules provided by packages.
All Univention Configuration Registry settings for filter rules are entered in the following format:
security/packetfilter/
protocol
/
port(s)
/
address
=
policy
security/packetfilter/package/
package
/
protocol
/
port(s)
/
address
=
policy
The following values need to be filled in:
package
(only for packaged rules)The name of the package providing the rule.
protocol
Can be either tcp
for server services using the Transmission Control Protocol or udp
for services using the stateless User Datagram Protocol.
port
,min-port
:
max-port
Ports can be defined either as a single number between 1 and 65535 or as a range separated by a colon:
min-port
:
max-port
address
This can be either ipv4
for all IPv4 addresses, ipv6
for all IPv6 addresses, all
for both IPv4 and IPv6 addresses, or any explicitly specified IPv4 or IPv6 address.
policy
If a rule is registered as DROP
, then packets to this port will be silently discarded;
REJECT
can be used to send back an ICMP message port unreachable
instead.
Using ACCEPT
explicitly allows such packets.
(IPtables rules are executed until one rule applies;
thus, if a package is accepted by a rule which is discarded by a later rule, then the rule for discarding the package does not become valid).
Filter rules can optionally be described by setting additional Univention Configuration Registry variables.
For each rule and language, an additional variable suffixed by “/
language
” can be used to add a descriptive text.
Some examples:
Example 14.1. Local firewall rule
security/packetfilter/tcp/2000/all=DROP security/packetfilter/tcp/2000/all/en=Drop all packets to TCP port 2000 security/packetfilter/udp/500:600/all=ACCEPT security/packetfilter/udp/500:600/all/en=Accept UDP port 500 to 600
All package rules can be globally disabled by setting the Univention Configuration Registry variable security/packetfilter/use_packages
to false.
.
iptables
commands
Besides the existing possibilities for settings via Univention Configuration Registry, there is also the possibility of integrating user-defined enhanced configurations in /etc/security/packetfilter.d/
, e.g. for realizing a firewall or Network Address Translation.
The enhancements should be realized in the form of shell scripts which execute the corresponding iptables
for IPv4 and ip6table
for IPv6 calls.
For packages this is best done through using a Univention Configuration Registry template as described in Section 2.2.1.1.
Full documentation for IPTables can be found at http://www.netfilter.org/.
Package filter settings should always be thoroughly tested.
The network scanner nmap
, which is integrated in Univention Corporate Server as a standard feature, can be used for testing the status of individual ports.
Since Nmap requires elevated privileges in the network stack, it should be started as root
user.
A TCP port can be tested with the following command:
nmap
HOSTNAME
-p
PORT(s)
A UDP port can be tested with the following command:
nmap
HOSTNAME
-sU
-p
PORT(s)
Example 14.2. Using nmap
for firewall port testing
nmap 192.168.1.100 -p 400 nmap 192.168.1.110 -sU -p 400-500
UCS is neither error free nor feature complete. Issues are tracked using Bugzilla at https://forge.univention.org/bugzilla/.
Create an account.
Search for existing entries before opening new reports.
Include the version info: ucr search --brief ^version/
.
Provide enough information to help us reproduce the bug.
Search http://sdb.univention.de/
Search http://wiki.univention.de/
Search http://forum.univention.de/ and ask for help. In addition to our support team many of our partners are also present there. Your questions might also help other users while you may profit from issues already solved for other users.
This chapter describes how software for Univention Corporate Server is packaged in the Debian format. It allows proper dependency handling and guarantees proper tracking of file ownership. Customers can package their own internal software or use the package mechanism to distribute configuration files consistently to different machines.
Software is packaged as a source package, from which one or more binary packages can be created.
This is useful to create different packages from the same source package.
For example the Samba source package creates multiple binary packages: one containing the file server, one containing the client commands to access the server, and several other packages containing documentation, libraries, and common files shared between those packages,
The directory should be named
.
package_name
-version
Some packages are required for creating and building packages.
This meta package depends on several other packages like compilers and tools to extract and build source packages. Packages must not declare an explicit dependency on this and its dependent packages.
This package contains additional scripts to modify source package files like for example debian/changelog
.
This program helps to create an initial debian/
directory, which can be used as a starting point for packaging new software.
These packages must be installed on the development system.
If not, missing packages can be installed on the command line using univention-install
or through UMC, which is described in the [ucs-handbuch].
dh_make
dh_make
is a tool, which helps creating the initial debian/
directory.
It is interactive by default and asks several questions about the package to be created.
Type of package: single binary, indep binary, multiple binary, library, kernel module, kernel patch? [s/i/m/l/k/n]
A single architecture specific binary package is created from the source package.
This is for software which needs to be compiled individually for different CPU architectures like i386
and amd64
.
A single architecture-independent binary package is created from the source package. This is for software which runs unmodified on all CPU architectures.
Multiple binary package are created from the source package, which can be both architecture independent and dependent.
Two or more binary packages are created for a compiled library package. The runtime package consists of the shared object file, which is required for running programs using that library. The development package contains the header files and other files, which are only needed when compiling and linking programs on a development system.
A single kernel-dependent binary package is created from the source package. Kernel modules need to be compiled for each kernel flavor. dkms should probably be used instead. This type of packages is not described in this manual.
A single kernel-independent package is created from the source package, which contains a patch to be applied against an unpacked Linux kernel source tree. dkms should probably be used instead. This type of packages is not described in this manual.
In Debian a package normally consists of an upstream software archive, which is provided by a third party like the Samba team.
This collection is extended by a Debian specific second TAR archive or a patch file, which adds the debian/
directory and might also modify upstream files for better integration into a Debian system.
When a source package is built, dpkg-source(1) separates the files belonging to the packaging process from files belonging to the upstream package.
For this to work, dpkg-source
needs the original source either provided as a TAR archive or a separate directory containing the unpacked source.
If neither of these is found and --native
is not given, dh_make
prints the following warning:
Could not find my-package_1.0.orig.tar.gz Either specify an alternate file to use with -f, or add --createorig to create one.
The warning from dh_make
states that no pristine upstream archive was found, which prohibits the creation of the Debian specific patch, since the Debian packaging tools have no way to separate upstream files from files specific to Debian packaging.
The option --createorig
can be passed to dh_make
to create a .orig.tar.gz
archive before creating the debian/
directory, if such separation is required.
The control files in the debian/
directory control the package creation process.
The following sections provide a short description of these files.
A more detailed description is available in the [Debian FAQ].
Several files will have the .ex
suffix, which mark them as examples.
To activate these files, they must be renamed by stripping this suffix.
Otherwise the files should be deleted to not clutter up the directory by unused files.
In case a file was deleted and needs to be restored, the original templates can be found in the /usr/share/debhelper/dh_make/debian/
directory.
The debian/
directory contains some global configuration files, which can be put into two categories:
The files changelog
, control
, copyright
, rules
are required and control the build process of all binary packages.
Most other files are optional and only affect a single binary package.
Their filename is prefixed with the name of the binary package,
[7]
The following files are required:
changelog
Changes related to packaging, not the upstream package. See Section B.3.3 below for more information.
compat
The Debhelper tools support different compatibility levels.
For UCS-3.x the file must contain a single line with the value 7
.
See debhelper(7) for more details.
control
Contains control information about the source and all its binary packages. This mostly includes package name and dependency informations. See Section B.3.1 below for more information.
copyright
This file contains the copyright and license information for all files contained in the package. See Section B.3.2 below for more information.
rules
This is a Makefile style file, which controls the package build process. See Section B.3.4 below for more information.
source/format
This file configures how dpkg-source(1) separates the files belonging to the packaging process from files belonging to the upstream package.
Historically the Debian source format 1.0
shipped packages as a TAR file containing the upstream source plus one patch file, which contained all files of the debian/
sub-directory in addition to all changes to upstream files.
The new format 3.0 (quilt)
replaces the patch file with a second TAR archive containing the debian/
directory.
Changes to upstream files are no longer applied as one giant patch, but split into logical changes and applied using a built-in quilt(1).
For simple packages, where there is no distinction between upstream and the packaging entity, the 3.0 (native)
format can be used instead, were all files including the debian/
directory are contained in a single TAR file.
The following files are optional and should be deleted if unused, which helps other developers to concentrate on only the files relevant to the packaging process:
README.Debian
Notes regarding package specific changes and differences to default options, for example compiler options.
Will be installed into /usr/share/doc/
.
package_name
/README.Debian
package
.cron.d
Cron tab entries to be installed. See dh_installcron(1) for more details.
package
.dirs
List of extra directories to be created. See dh_installdirs(1) for more details. [8]
package
.install
List of files and directories to be copied into the package. This is normally used to partition all files to be installed into separate packages, but can also be used to install arbitrary files into packages. See dh_install(1) for more details.
package
.docs
List of documentation files to be installed in /usr/share/doc/
.
See dh_installdocs(1) for more details.
package
/
package
.emacsen-install
,package
.emacsen-remove
,package
.emacsen-startup
Emacs specific files to be installed below /usr/share/emacs-common/
.
See dh_installemacsen(1) for more details.
package
/
package
.doc-base*
Control files to install and register extended HTML and PDF documentation. See dh_installdocs(1) for more details.
package
.init.d
,package
.default
Start-/stop script to manage a system daemon or service. See dh_installinit(1) for more details.
package
.manpage.1
,package
.manpage.sgml
Manual page for programs, library functions or file formats, either directly in troff or SGML. See dh_installman(1) for more details.
package
.menu
Control file to register programs with the Debian menu system. See dh_installmenu(1) for more details.
watch
Control file to specify the download location of this upstream package. This can be used to check for new software versions. See uscan(1) for more details.
package
.preinst
,package
.postinst
,package
.prerm
,package
.postrm
Scripts to be executed before and after package installation and removal. See Section B.3.5 below for more information.
package
.maintscript
Control file to simplify the handling of conffiles. See dpkg-maintscript-helper(1) and dh_installdeb(1) for more information.
Other debhelper programs use additional files, which are described in the respective manual pages.
The control
file contains information about the packages and their dependencies, which is needed by dpkg
.
The initial control
file created by dh_make
looks like this:
Source: testdeb Section: unknown Priority: optional Maintainer: John Doe <user@example.com> Build-Depends: debhelper (>= 5.0.0) Standards-Version: 3.7.2 Package: testdeb Architecture: any Depends: ${shlibs:Depends}, ${misc:Depends} Description: <insert up to 60 chars description> <insert long description, indented with spaces>
The first block beginning with Source
describes the source package:
Source
The name of the source package.
Must be consistent with the directory name of the package and the information in the changelog
file.
Section
A category name, which is used to group packages.
There are many predefined categories like libs
, editors
, mail
, but any other string can be used to define a custom group.
Priority
Defines the priority of the package. This information is only used by some tools to create installation DVD. More important packages are put on earlier CD, while less important packages are put on later CD.
essential
Packages are installed by default and dpkg
prevents the user from easily removing it.
required
Packages which are necessary for the proper functioning of the system. The package is part of the base installation.
important
Important programs, including those which one would expect to find on any Unix-like system. The package is part of the base installation.
standard
These packages provide a reasonably small but not too limited character-mode system.
optional
Package is not installed by default. This level is recommended for most packages.
extra
This contains all packages that conflict with some other packages.
Maintainer
The name and email address of a person or group responsible for the packaging.
Build-Depends
,Build-Depends-Indep
A list of packages which are required for building the package.
Standards-version
Specifies the Debian Packaging Standards version, which this package is conforming to. This is not used by UCS, but required by Debian.
All further blocks beginning with Package
describes a binary package.
For each binary package one block is required.
Package
The name of the binary package. The name must only consist of lower case letters, digits and dashes. If only a single binary package is build from a source package, the name is usually the same as the source package name.
Architecture
Basically there are two types of packages:
Architecture dependent packages must be build for each architecture like i386
and amd64
, since binaries created on one architecture do not run on other architectures.
A list of architectures can be explicitly given, or any
can be used, which is then automatically replaced by the architecture of the system where the package is built.
Architecture independent packages only need to be built once, but can be installed on all architectures.
Examples are documentation, scripts and graphics files.
They are declared using all
in the architecture field.
Description
The first line should contain a short description of up to 60 characters, which should describe the purpose of the package sufficiently. A longer description can be given after that, where each line is indented by a single space. An empty line can be inserted by putting a single dot after the leading space.
Most packages are not self-contained but need other packages for proper function. Debian supports different kinds of dependencies.
Depends
A essential dependency on some other packages, which must be already installed and configured before this package is configured.
Recommends
A strong dependency on some other packages, which should normally be co-installed with this package, but can be removed. This is useful for additional software like plug-ins, which extends the functionality of this package, but is not strictly required.
Suggests
A soft dependency on some other packages, which are not installed by default. This is useful for additional software like large add-on packages and documentation, which extends the functionality of this package, but is not strictly required.
Pre-Depends
A strong dependency on some other package, which must be fully operational even before this package is unpacked.
This kind of dependency should be used very sparsely.
It's mostly only required for software called from the .preinst
script.
Conflicts
A negative dependency, which prevents the package to be installed while the other package is already installed. This should be used for packages, which contain the same files or use the same resources, for example TCP port numbers.
Provides
This package declares, that it provides the functionality of some other package and can be considered as a replacement for that package.
Replaces
A declaration, that this package overwrites the files contained in some other package.
This deactivates the check normally done by dpkg
to prevent packages from overwriting files belonging to some other package.
Breaks
A negative dependency, which requests the other package to be upgraded before this package can be installed.
This is a lesser form of Conflicts
.
Breaks
is almost always used with a version specification in the form Breaks:
:
This forces package
(<< version
)package
to be upgraded to a version greater than version
before this package is installed.
In addition to literal package names, debhelper supports a substitution mechanism: Several helper scripts are capable of automatically detecting dependencies, which are stored in variables.
${shlibs:Depends}
dh_shlibdeps
automatically determines the shared library used by the programs and libraries of the package and stores the package names providing them in this variable.${python:Depends}
dh_python
detects similar dependencies for Python modules.${misc:Depends}
In addition to specifying a single package as a dependency, multiple packages can be separated by using the pipe symbol (|
).
At least one of those packages must be installed to satisfy the dependency.
If none of them is installed, the first package is chosen as the default.
A package name can be followed by a version constraint enclosed in parenthesis. The following operators are valid:
<<
<=
=
>=
>>
An Example:
Depends: libexample1 (>= ${binary:Version}), exim4 | mail-transport-agent, ${shlibs:Depends}, ${misc:Depends} Conflicts: libgg0, libggi1 Recommends: libncurses5 (>> 5.3) Suggests: libgii0-target-x (= 1:0.8.5-2) Replaces: vim-python (<< 6.0), vim-tcl (<= 6.0) Provides: www-browser, news-reader
The copyright
file contains copyright and license information.
For a downloaded source package it should include the download location and names of upstream authors.
This package was debianized by John Doe <max@example.com> on Mon, 21 Mar 2009 13:46:39 +0100. It was downloaded from <fill in ftp site> Copyright: Upstream Author(s): <put author(s) name and email here> License: <Must follow here>
The file does not require any specific format. Debian now recommends to use a machine-readable format, but this is not required for UCS. The format is described in http://dep.debian.net/deps/dep5/ at looks like this:
Format: http://www.debian.org/doc/packaging-manuals/copyright-format/1.0/ Upstream-Name: Univention GmbH Upstream-Contact: <package@univention.de> Source: https://docs.software-univention.de/ Files: * Copyright: 2013-2016 Univention GmbH License: AGPL
The changelog
file documents the changes applied to this Debian package.
The initial file created by dh_make
only contains a single entry and looks like this:
testdeb (0.1-1) unstable; urgency=low * Initial Release. -- John Doe <user@example.com> Mon, 21 Mar 2013 13:46:39 +0100
For each new package release a new entry must be prepended before all previous entries. The version number needs to be incremented and a descriptive text should be added to describe the change.
The command debchange
from the devscripts package can be used for editing the changelog
file.
For example the following command adds a new version:
dch -i
After that the changelog
file should look like this:
testdeb (0.1-2) unstable; urgency=low * Add more details. -- John Doe <user@example.com> Mon, 21 Mar 2013 17:55:47 +0100 testdeb (0.1-1) unstable; urgency=low * Initial Release. -- John Doe <user@example.com> Mon, 21 Mar 2013 13:46:39 +0100
The date and time stamp must follow the format described in RFC 2822.
debchange
automatically inserts and updates the current date.
Alternatively date -R
can be used on the command line to create the correct format.
For UCS it is best practice to mention the bug ID of the UCS bug tracker (see Appendix A) to reference additional details of the bug fixed. Other parties are encouraged to devise similar comments, e.g. URLs to other bug tracking systems.
The file rules
describes the commands needed to build the package.
It must use the Make syntax [make].
It consists of several rules, which have the following structure:
target: dependencies command ...
Each rule starts with the target name, which can be a file name or symbolic name. Debian requires the following targets:
clean
build
,build-arch
,build-indep
binary
,binary-arch
,binary-indep
debian/tmp/
below the source package root directory.
From there files are distributed to individual packages, which are created as the result of these rules.
These rules are called with root permissions.
Each command line must be indented with one tabulator character.
Each command is executed in a separate shell, but long command lines can be split over consecutive lines by terminating each line with a backslash (\
).
Each rule describes a dependency between the target and its dependencies.
make
considers a target to be out-of-date, when a file with that name target
does not exists or when the file is older than one of the files it depends on.
In that case make
invokes the given commands to re-create the target.
In addition to file names also any other word can be used for target names and in dependencies.
This is most often used to define “phony” targets, which can be given on the command line invocation to trigger some tasks.
The above mentioned clean
, build
and binary
targets are examples for that kind of targets.
dh_make
only creates a template for the rules
file.
The initial content looks like this:
#!/usr/bin/make -f # -*- makefile -*- # Sample debian/rules that uses debhelper. # This file was originally written by Joey Hess and Craig Small. # As a special exception, when this file is copied by dh-make into a # dh-make output file, you may use that output file without restriction. # This special exception was added by Craig Small in version 0.37 of dh-make. # Uncomment this to turn on verbose mode. #export DH_VERBOSE=1 %: dh $@
Since UCS-3.0 the debian/rules
file is greatly simplified by using the dh
sequencer.
It is a wrapper around all the different debhelper tools, which are automatically called in the right order.
To exactly see which commands are executed when dpkg-buildpackage
builds a package, invoke dh
by hand, for example target
--no-actdh binary --no-act
lists all commands to configure, build, install and create the package.
In most cases it's sufficient to just provide additional configuration files for the individual debhelper commands as described in Section B.3.
If this is not sufficient, any debhelper command can be individually overridden by adding an override target to the rules
file.
For example the following snippet disables the automatic detection of the build system used to build the package and passes additional options:
override_dh_auto_configure: ./setup --prefix=/usr --with-option-foo
Without that explicit override dh_auto_configure
would be called, which normally automatically detects several build systems like cmake, setup.py, autoconf and others. For these dh
also passes the right options to configure the default prefix /usr
and use the right compiler flags.
After configuration the package is built and installed to the temporary staging area in debian/tmp/
.
From there dh_install
partitions individual files and directories to binary packages.
This is controlled through the debian/
files.
package
.install
This file can also be used for simple packages, where no build system is used.
If a path given in the debian/
file is not found below package
.installdebian/tmp/
, the path is interpreted as relative to the source package root directory.
This mechanism is sufficient to install simple files, but fails when files must be renamed or file permissions must be modified.
In addition to distributing only files packages can also be used to run arbitrary commands on installation, upgrades or removal. This is handled by the four “Maintainer scripts”, which are called before and after files are unpacked or removed:
debian/package
.preinst
debian/package
.postinst
debian/package
.prerm
debian/package
.postrm
The scripts themselves must be shell scripts, which should contain a #DEBHELPER#
marker, where the shell script fragments created by the dh_
programs are inserted.
Each script is invoked with several parameters, from which the script can determine, if the package is freshly installed, upgraded from a previous version, or removed.
The exact arguments are described in the template files generated by dh_make
.
The maintainer scripts can be called multiple times, especially when errors occur. Because of that the scripts should be idempotent, that is they should be written to “achieve a consistent state” instead of blindly doing the same sequence of commands again and again. A bad example would be to append some lines to a file on each invocation. The right approach would be to add a check, if that line was already added and only do it otherwise.
It is important that these scripts handle error conditions properly:
Maintainer scripts should terminate with exit 0
on success and exit 1
on fail, if things go catastrophically wrong.
On the other hand an exit code unequal to zero usually aborts any package installation, upgrade or removal process. This prevents any automatic package maintenance and usually requires manual intervention of a human administrator. Therefore it is essential that maintainer scripts handle error conditions properly and are able to recover an inconsistent state.
Before the first build is started, remove all unused files from the debian/
directory.
This simplifies maintenance of the package and helps other maintainers to concentrate on only the relevant differences from standard packages.
The build process is started by invoking the following command:
dpkg-buildpackage -us -uc
The options -us
and -uc
disable the PGP signing process of the source and changes files.
This is only needed for Debian packages, were all files must be cryptographically signed to be uploaded to the Debian infrastructure.
Additionally the option -b
can be added to restrict the build process to only build the binary packages.
Otherwise a source package will also be created.
[ucs-handbuch] Univention GmbH. 2018. Univention Corporate Server - Manual for users and administrators. https://docs.software-univention.de/manual-4.3.html.
[make] Free Software Foundation. 2010. The GNU Make manual.
[ISO639] International Organization for Standardization. 2002. ISO 639-1: Alpha-2 code.
[Debian FAQ] Debian. 2012. The Debian GNU/Linux FAQ - Basics of the Debian package management system.
[Debian Guide] Debian. 2013. Debian New Maintainers' Guide.
[Debian Policy] Debian. 2012. Debian Policy Manual.
[Debian Reference] Debian. 2012. Debian Developer's Reference.