#555 change "apache::mod::passenger" to support overriding all parameters expected by "apache::mod". #558
Conversation
The current code allows you to set index_options but then ignores it.
Based on the git log for the commit that added this option it was
intended to be passed into directories. The test that caught this
will be merged into spec/acceptance/vhost_spec.rb in a seperate PR.
You can reproduce the original failure with:
class { 'apache': }
host { 'test.server': ip => '127.0.0.1' }
apache::vhost { 'test.server':
docroot => '/tmp',
index_options => ['Charset=UTF-8'],
}
We're just going to remove them because they can be passed directly
into directory instead of indirectly.
The servername can be [schema://]hostname.domain[:port]. In the case of servername including schema and port, it is not appropriate to use as log file name. It will generate something like this: "ErrorLog /var/log/httpd/https://domain_error.log", which cause httpd failed to refresh/start.
…od::passenger "apache::mod::passenger" was not accepting/passing (optional) mod paramters to apache::mod, preventing overriding the auto-generated module parameters. Without this option the apache::mod::passenger could not be used in conjunction with passenger modules generated outside the system package system (in particular, when compiled via passenger-install-apache2-module), as those are output to different paths). This patch enables passing in all parameters used with "apache::mod".
support apache::mod params added to apache::mod:passenger with specs (coverage is currently ubuntu-specific, along with the rest of the spec coverage)
|
This will need spec/acceptance tests, and documentation. |
|
I've got docs ready and spec/acceptance tests working for ubuntu (testing with the beaker set-up). However, HEL/Centos/SL's mod_passenger appears to have long-unstated dependencies on two separate 3rd-party repos; one is Passenger-specific and the other is EPEL . Any opinions on embedding these requirements in the module vs. pre-installing those dependencies "just for testing"? If you prefer these to go into the module, are there any concerns with adding a dependency on the stahnma/epel 3rd-party module? |
add basic spec/acceptance tests for passenger module; these verify a default installation as well as passing in module loading parameters. new tests revealed undocumented Redhat dependencies on passenger- specific repository as well as the EPEL repository; as a first step these dependencies are being manually resolved during setup for the acceptance testing suite. also, documented new apache::mod::passenger parameters (originatinig with apache::mod) in README.passenger.md
This is implemented quite similar to RequestHeader directives support. Includes updated documentation and a simple spec test. Fixes issue #573.
…'t lost The mod_passenger RPM supplies /etc/httpd/conf.d/passenger.conf with the correct PassengerRoot for the version of the package installed. The most reliable way to keep the PassengerRoot accurate is to leave this file in place and then to install a second file (passenger_extra.conf) for other Passenger customisations. Hardcoding the PassengerRoot in the module means it gets out of step with package repositories (3.0.21 is in EPEL6 at the time of writing, while the module assumes .17). Passenger 4.x refuses to start if the PassengerRoot is incorrect. The user of apache::mod::passenger is also able to override this value, but they shouldn't be forced to, just to get a functioning install. Fixes #560
|
Are there any further updates that would be useful to make here? |
Delete this key, mistakenly added.
Pass this through to directories.
Configure Passenger in separate .conf file on RH so PassengerRoot isn't lost
added redirectmatch support
Support Header directives in vhost context
Remove spec:system references and remove tabs from docs
Fix the servername used in log file name
|
Sorry for sleeping on the job! Apparently you'll have to rebase against the current master |
Don't purge mods-available dir when separate enable dir is used
…od::passenger "apache::mod::passenger" was not accepting/passing (optional) mod paramters to apache::mod, preventing overriding the auto-generated module parameters. Without this option the apache::mod::passenger could not be used in conjunction with passenger modules generated outside the system package system (in particular, when compiled via passenger-install-apache2-module), as those are output to different paths). This patch enables passing in all parameters used with "apache::mod".
support apache::mod params added to apache::mod:passenger with specs (coverage is currently ubuntu-specific, along with the rest of the spec coverage)
add basic spec/acceptance tests for passenger module; these verify a default installation as well as passing in module loading parameters. new tests revealed undocumented Redhat dependencies on passenger- specific repository as well as the EPEL repository; as a first step these dependencies are being manually resolved during setup for the acceptance testing suite. also, documented new apache::mod::passenger parameters (originatinig with apache::mod) in README.passenger.md
rebase to master and fix following conflicts: - correct apache::params on redhat systems for passenger_root and passenger_ruby to ensure a working default configuration - fix expected paths in mod_passenger_spec to accomodate system-specific passenger configuration paths Also: - add notice of redhat-specific package repository dependencies to passenger README
|
No worries, thanks for the heads-up |
correct passenger specs for RedHat - these were out-of-date with the revised default passenger params (for RedHat)
|
I've committed a fix for the spec/classes/passenger_spec.rb failures from the Travis CI build |
|
If this needs to be rebased to master, could you recommend working versions for serverspec and specinfra, etc? I'm currently seeing 100% failure when running spec/acceptance tests against the latest versions. Thanks! |
|
Just saw the pending pull request in beaker-rspec; I'll wait on that fix. :) |
|
Looks like the rebase to master didn't quite work. This PR still has 24 commits and a bunch of merge commits in it. |
|
Should I re-do the changes in a new branch off master and create a new PR? I'm fine with whatever approach will be easiest for reviewing (esp. now that beaker-rspec is fixed and I can run acceptance tests). |
|
@jonoterc whatever is easier to you. We're quite flexible! |
|
Hi, it seemed simplest to cleanly apply the updates to a new branch off an up-to-date master - I've submitted a new PR: #607 |
|
Ok, I'll close this one then. Thanks! |
(maint) Release Mergeback version v3.9.0
apache::mod::passenger did not accept/pass (optional) module parameters to apache::mod,
thus preventing overriding the auto-generated module parameters. Without this option
apache::mod::passenger could not be used in conjunction with passenger modules generated
outside the OS package system (in particular, when compiled via
passenger-install-apache2-module), as those are output to different paths. This patch
enables overriding all parameters expected by "apache::mod", and includes spec coverage.