Manages Selenium Server 2 - Standalone and Grid
This is a puppet module for installation and configuration of the Selenium
Server 2 browser
automation package. Support is provided for configuring Selenium Server in
standalone mode or as a Selenium grid with composed of the hub and node
modes.
The design intent of this module is to only manage Selenium server and not any of the other packages that might be required to required to create a selenium server profile/role.
The selenium::server class is used to setup a standalone Selenium instance
to allow the use of a single server as a test node. The selenium::hub class
acts as a proxy in front of one or more selenium::node instances. A hub +
node(s) setup is referred to as a Selenium grid. Running selenium::server is
similar to creating a Selenium grid by declaring selenium::hub and
selenium::node on the same host.
It is highly recommend that you read the Selenium Grid documentation to determine what the best approach for your usage case(s) is.
Stand alone server setup with display :99 (default is :0).
class { 'selenium::server': display => ':99' }Grid in a box setup with the Selenium node talking to the hub via
127.0.0.1.
class { 'selenium::hub': }
class { 'selenium::node': }Here are some examples "profiles" and "roles" based on Craig Dunn's blog post on Designing Puppet – Roles and Profiles..
These examples assume the presence of these two modules in your Puppet environment.
class mysite::profile::seleniumserver {
include java
# WSXGA+ 1680x1050 -- should nicely fit on a 1920x1280 screen
class { 'display':
width => 1680,
height => 1050,
} ->
class { 'selenium::server': }
Class['java'] -> Class['selenium::server']
}class mysite::profile::seleniumhub {
include java
class { 'selenium::hub': }
Class['java'] -> Class['selenium::hub']
}class mysite::profile::seleniumnode {
include java
# WSXGA+ 1680x1050 -- should nicely fit on a 1920x1280 screen
class { 'display':
width => 1680,
height => 1050,
} ->
class { 'selenium::node':
# If your intending to have node(s) that don't sit on the same system as
# the hub, you need to point the node at a hub by passing in the hub's url
# in or using an exported resource from the hub system. Eg.
# hub => 'http://<myseleniumhub>:4444/grid/register',
}
Class['java'] -> Class['selenium::node']
}
class mysite::role::seleniumserver {
include mysite::profile::seleniumserver
}class mysite::role::seleniumhub {
include mysite::profile::seleniumhub
include mysite::profile::seleniumnode
}Unless class selenium has $install_root changed from the default, these
files paths will be used. Note that the log files and init scripts are only
created for the relevant configured service.
The log files are set to mode 0644, so that they can be inspected by users
other than the selenium role account.
/opt/selenium
|-- jars
| `-- selenium-server-standalone-x.xx.x.jar
`-- log
|-- hub_stderr.log
|-- hub_stdout.log
|-- node_stderr.log
|-- node_stdout.log
|-- server_stderr.log
`-- server_stdout.log
/var/log/selenium -> /opt/selenium/log
/etc/init.d/seleniumhub
/etc/init.d/seleniumnode
/etc/init.d/seleniumserver
This class controls common configuration values used by the
selenium::{server,hub,node} classes. It is automatically included in the
manifest by those classes and it need not be explicitly declared except to
override the default values.
# defaults
class { 'selenium':
user => 'selenium',
manage_user => true,
group => 'selenium',
manage_group => true,
install_root => '/opt/selenium',
java => 'java',
version => '2.45.0',
url => undef,
download_timeout => '90',
nocheckcertificate => false,
}String defaults to: selenium
The name/uid of the system role account to execute the server process under and will have ownership of files.
Boolean defaults to: true
Whether or not this module should manage the system role account to execute the server process under.
String defaults to: selenium
The group/gid of the system role account and group ownership of files.
Boolean defaults to: true
Weather or not this module should manage the group of the system role account.
String defaults to: /opt/selenium
The dirname under which Selenium Server files (including logs) will be created.
String defaults to: java
The path of the java interpreter to use.
String defaults to: 2.45.0 (the latest release as of 2015-05-23)
The version of Selenium Server to download. Used to form the URL used to fetch the jar file.
String defaults to: undef
If defined, url will be used to download the Selenium Server jar file.
However, the version parameter just match the version of the downloaded file
as this information is needed when starting up the server (this may change to
be be automatically parsed from the url in a later release).
String defaults to: 90
Timeout to download of the package.
Boolean defaults to: false
Disables validation of the x509 certificate the Selenium jar file is retrieved from.
# defaults
class { 'selenium::server':
display => ':0',
options => '-Dwebdriver.enable.native.events=1',
}String defaults to: :0
The name of the X display to render too. This is set as an environment
variable passed to Selenium Server
String defaults to: -Dwebdriver.enable.native.events=1
Options passed to Selenium Server at startup.
Note that by default selenium::server and selenium::hub will try to listen
on the same TCP port (4444) and only one of them will be able to function.
# defaults
class { 'selenium::hub':
options => '-role hub',
}String defaults to: -role hub
Options passed to Selenium Server Hub at startup.
# defaults
class { 'selenium::node':
display => ':0',
options => '-Dwebdriver.enable.native.events=1 -role node',
hub => 'http://localhost:4444/grid/register',
}String defaults to: :0
The name of the X display to render too. This is set as an environment
variable passed to Selenium Server
String defaults to: -Dwebdriver.enable.native.events=1 -role node
Options passed to Selenium Server Node at startup.
String defaults to: http://localhost:4444/grid/register
The URL of the Selenium Server Hub to connect to.
At the time work on this module was started, there were no other Selenium Server modules published on the Puppet Forge. A number of existing modules were identified on github but none of them fit the author's needs of:
- Allowing the external setup of things like
Xvfb,java, etc. and to allow the composition of site specific profiles/roles. - Supporting a mixed environment of Scientific, Centos, and RedHat Linux
(basically
$::osfamily == 'RedHat' - Enough
rspec-puppetcoverage to prevent regressions
The later is the most important issue. Attempting to add additional
$::operatingsystem/$::osfamily support to a puppet module without
rspec-puppet tests is a process fraught with peril.
The modules that were identified were:
- adamgoucher/selenium-puppet
- Mostly aimed at Windows but with some Linux/POSIX support
- no
rspec-puppettests (deal breaker)
- StoryIQ/puppet-selenium-grid
- Debian only
- no
rspec-puppettests (deal breaker)
- kayakco/puppet-selenium
- Unreleased deps (deal breaker)
- Very role like with lots of magic setup (deal breaker)
- debian-ish (may support Centos)
- Doesn't appear to be in a ready state
- el6.x
- el7.x
- Debian 7
- Ubuntu 12.04
- Ubuntu 14.04
| Versions | Puppet 2.7 | Puppet 3.x | Puppet 4.x |
|---|---|---|---|
| 0.x | yes | yes | no |
| 1.x | no | yes | yes |
A number of users have reported "certificate validation errors" when this
module downloads the Selenium jar file. The most likely explanation is that
the CA certificates on that node are out of date. It may also be an
indication of a MITM attack on the TLS connection. Certificate validation
errors do not occur on the nodesets used for acceptance tests. The
nocheckcertificate param may be set to true to
bypass this error but be sure the security implications of this setting are
well understood before enabling it.
This module is versioned according to the Semantic Versioning 2.0.0 specification.
Please log tickets and issues at github
- Fork it on github
- Make a local clone of your fork
- Create a topic branch. Eg,
feature/mousetrap - Make/commit changes
- Commit messages should be in imperative tense
- Check that linter warnings or errors are not introduced -
bundle exec rake lint - Check that
Rspec-puppetunit tests are not broken and coverage is added for new features -bundle exec rake spec - Documentation of API/features is updated as appropriate in the README
- If present,
beakeracceptance tests should be run and potentially updated -bundle exec rake beaker
- When the feature is complete, rebase / squash the branch history as necessary to remove "fix typo", "oops", "whitespace" and other trivial commits
- Push the topic branch to github
- Open a Pull Request (PR) from the topic branch onto parent repo's
masterbranch
