-
Notifications
You must be signed in to change notification settings - Fork 1.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fixes/apache name #1070
Fixes/apache name #1070
Conversation
ok... now I'm lost In the Travis log: $concat_basedir not defined. Try running again with pluginsync=true on the [master] and/or [main] section of your node's '/etc/puppet/puppet.conf'. at /home/travis/build/puppetlabs/puppetlabs-apache/spec/fixtures/modules/concat/manifests/setup.pp:20 My local test run was fine. |
My Local run was againt another branch... stupid me... This 'dev' class seems broken to me, it eiter requires apache::package(FreeBSD) or apache, but not both, as that causes a duplicate resource error. |
|
2944add
to
2221328
Compare
Ok, I got it down again to 1 error: FreeBSD So... I'm stuck now. |
I'm not all that familiar with FreeBSD, but is that actually needed? |
Not 100% sure, but it looks like the purpose of The installation of apache package(s) is handled quite differently under FreeBSD. You may skim through a comment in init.pp to get some insight. In most systems the list of |
2221328
to
4a1bce1
Compare
I removed the 'fail()' condition for FreeBSD, this allows all tests to pass. Also, I don't really like how the module is structured so differently between FreeBSD and other systems, to be honest. I think it should be more uniform, either moving the install for FreeBSD back to the init class like the others, or vice versa. Does FreeBSD allow the complete Apache httpd package to be compiled like other systems do? Where the different mpm modules are compiled anyway, and use configuration to load the required module? I guess it would make things simpler. |
It was a huge changeset when I was adding the FreeBSD support, and I decided to "not touch" other systems (even if this could unify code). I was almost sure, that it would not be accepted if I altered other OSes. But FreeBSD was really specific...
Using special manifest, such as the
Actually, it's a standard approach on FreeBSD -- to compile source packages (ports) instead of just install binary packages from a repository. It was an approach, at least at that time I was struggling with FreeBSD (8.X). I'm not sure how this evolved since that.
For apache 2.2 there were separate FreeBSD ports for different MPMs. For example, in order to use ITK MPM, you had to compile and install a port named |
99e79ee
to
c0aff21
Compare
We're making progress... this is the second test run that failed in 'strict' due to some fact not being available in the test. Now running the third... |
c0aff21
to
d506c03
Compare
@stevenpost could you squash your three commits into one? Thanks! |
This also implies the dependency for the tests needed to be changed. Because of the changed dependency, tests needed some extra facts to please the concat module.
d506c03
to
296aab5
Compare
|
Builds seem to fail due to availability problems with Github. |
thank you @stevenpost and @fraenki! |
Based on the work of @fraenki in #1056
His patch is rebased against a more recent master, and I added 2 patches.
This was not the case during the previous versions, which caused some tests to fail after changing the parameter references used.