pysa - Reverse Engineer Server Configurations
pysa [ -hqps ] [ -m module-name ] [ -o output-path ][-c config-file-path] [ -f filter-config-path ]
pysa scans your system and reverse engineers its configurations for easy replication.
pysa is able to scan your system, looking for different resources to deploy and generates some "autoconf" tools script to deploy it later on another computer.
See RESOURCES section for complete list of managed resources.
pysa is able to generates the configuration in Puppet (see Puppet documentation) or SaltStack (see SaltStack documentation) format.
Display the short help.
Activate quiet mode and displays only error messages. By default, pysa displays all log messages.
Generates Puppet output.
Generates SaltStack output.
Choose output module name. Default value: pysa
Specify a configuration file. See examples file for more details pysa.cfg
Choose the output filter for generated scripts. Default value: output
Specify a filter configuration file. See FILTERS_ section for more details.
pysa generates a puppet module containing several configuration scripts.
There are two ways to use pysa 's output:
- You can manually configure the configuration manager and add pysa 's module to it
- If you're using Puppet module, you can use the pysa2puppet script to deploy a complete and standalone setup. The script is interactive and will ask you all necessaries info (see usage first). A SaltStack version will be published soon.
This section describes all the resources scanned by pysa
By default, all item described are scanned. However, you can apply some filters to avoid or specify some. See the FILTERS section.
At the current state, the resources objects and keys are similar to Puppet types. Jump to pysa/scanner/object/ for a compelte object description. These objects will be documented soon.
Please see AUTOCONF TOOLS MODULES section to be sure to be able to handle all scanned resources.
Pleese note that in main cases, the scan must be done by an admin user (mostly root).
pysa scans (and stores in output module) all files located in a specific location. Default /etc and /root/.ssh
Primary key: path
pysa is able to scan all packages provided by yum, apt-get, python pypi (pip), ruby gems, nodejs packaged modules (npm) and php packages managers (pear and pecl).
Furthermore, pysa is able to detect repositories rpm packages if yum is not present.
Primary key: name
pysa detects all startup services managed by upstart and SysV init scripts.
Please see NOTES section.
Primary key: name
pysa scans and reproduces existing hostname associations (default /etc/hosts).
Primary key: name
pysa scans and reproduces existing users (/etc/passwd).
Primary key: name
pysa scans and reproduces existing groups (/etc/groups).
Primary key: name
pysa scans and reproduces existing mount points (/etc/fstab).
Primary key: device
pysa scans and reproduces user's crons.
Primary key: name
pysa scans and reproduces root SSH keys (default /root/.ssh).
SSH keys are manages as files.
Primary key: name
pysa is able to recognize all source repositories managed by the most common SCM (subversion, git and mercurial) present in the system.
Primary key: path
Puppet only The sources scanner is not able to scan sources repositories for SaltStack yet.
pysa scans and reproduces yum and apt-get repositories.
Primary key: name
This section lists the autoconf tools' modules which may be used.
Modules are used for particular features and are only needed in some particular cases. Of course, modules (as with the autoconf tools) have to be installed on the new machine, not the original one.
willdurand/nodejs: add npm package manager support nodes/php: add php package manager support puppetlabs/vcsrepo: add scm (sources) support
to install a Puppet module: puppet module install module-name
pysa integrates a powerful filters engine, which allows you to adapt its behavior to your needs.
A filter file is composed of sections, keys and values. In some specific cases sections and/or keys can be split using a '.' (see below for more details).
A key can be tagged with '_' at the front to be considered as "action" key. An action key is a key representing a specific action in the section (see below).
If some parameters conflict then the result may be harmful, please use it carefully. Don't hesitate to report any abnormal output to us.
Some improvements are planned in this section.
This key acts as a pointer. All the content of the referred section will be interpreted in the section.
This key should be set alone, as all keys will be replaced.
This section is used to add or modify some values.
It can sounds similar to the replace section, but works in a completely different way:
- The key is based on section key instead of content to replace
- It is replaced at the scanning step, while the replacement section is done at the output generation step
Remember that addition is used to add/set a concrete parameter, while replace is used to replace some content.
The section name can be separate in multiple subsections using a dot '.', always starting by addition keyword:
- addition.resource_type will replace values for all objects of resource_type
- addition.resource_type.key.value will replace only the values for the objects where the key/value match the requirement
The key represents the resource key. The value represents the resource value.
section_key = section_value
No action key for this section.
This section is used to specify which object should or shouldn't be discard.
The key is separated in to two sub-keys by a dot '.', which represents the object type for the first one and the attribute name for the second one.
The values can be seen as a list of attributes separated by a coma ','.
The joker '*' can be used to specify to match all characters.
object.attribute_name = attribute1, attribute2*, ...
- _resources:
- resource names Select which resources to be scanned, use it carefully, some resources might depend on others.
This section is used to replace any kind of content.
The section name can be separated into multiple subsections using a dot '.', always starting by replace keyword:
- replace will replace all values for all objects.
- replace.object will replace all values for the selected object.
- replace.object.field will replace only the values associated with the field in the selected object.
The key represents the new value. The value(s) represents the target to replace.
new_value = old_value1, old_value2, ...
_replaceall:
- true/false
- REQUIRED
- Select the filtering mode (replace all except -true- or replace none except -false-)
- default: true _except: primary_keys_values
This section is used to specify which package should be updated. This section has been created due to the lack of old packages in many repositories.
A list of package names is specified as values of the except key, separated by a coma ','.
The joker '*' can be used to specify to match all characters.
except = package1, package2*, *package3, *package4*, ...
_update:
- true/false
- REQUIRED
- Select the filtering mode (update all except -true- or update none except -false-)
- default: false
See docs/examples for configuration file examples.
pysa has been inspired by a software called Blueprint (more information at http://devstructure.com/blueprint/).
The force of pysa lies on the following points:
- pysa's "filters" and Blueprint's "rules" are totally different. Please refer to the documentations for more details.
- pysa's Puppet output is cleaner (the files are separated, the module is automatically created...).
- The dependency cycle is more resilient. pysa generates an attribute-based dependency cycle (each object relies and depends on its own dependencies) so if something fails the whole process isn't stopped.
- pysa is under active development and there is additional functionality under development (e.g., integration to Madeira's services, Salt/Chef modules).
As an early-release, pysa does not (always) provide 100% functional results. This comes, in some cases, from the architectural choices that we've made. For example, pysa does not (yet) support the addition of user's packages, simply because we can't ensure the availability of these packages on the new system. It would lead to the generation of wrong output files.
Furthermore, pysa depends on "autoconf" tools. This means that if a feature is not supported by one of these tools, pysa can't provide it. For example, it is currently impossible to use upstart services on a Redhat based platform, as it is impossible to use npm package manager on Ubuntu.
Please don't hesitate to contact us for any kind of feedback, advice or requirement: [email protected] for public discussions and [email protected] for private messages.
If you have a question about a specific source file, you can also contact the author directly ([email protected])
No known bugs.