Fix breaking change regression of PR #429#434
Conversation
Adding version constraint for the effectiveness of the change_dbpassword parameter to greater than 10.3.0. In all versions prior to 10.3.0 the password is written in plain text.
|
I really tried to add an acceptance test for JIRA 10.3.3 but I think we would need to limit those to certain modern OS versions like CentOS 9. Otherwise the prerequisites on MySQL and JAVA Version would not be able to be met. Unfortunately, I don't know how to limit the acceptance tests to a certain OS. If someone knows, please contact me. |
|
@diLLec looks good to me. I asked for two other reviewers |
|
@diLLec can you sqash your commits so there is just one? I think most commits can be fixups.
|
|
@diLLec thanks for the PR. As you mentioned, I think an acceptance test makes sense. https://github.com/voxpupuli/puppet-nodejs/blob/master/spec/acceptance/class_spec.rb#L38 has an example of limiting a test to specific operating systems. Does that help you? |
kenyon
left a comment
There was a problem hiding this comment.
We can squash commits on merge.
|
The title of this pull request ends up in the change log. |
|
There is another regression as well. when provisioning jira 10.3.x on a fresh server without specifying change_dbpassword = true, the password ends up being set to Solving this regression requires implementing some facts to figure out if jira is already installed or not. |
|
After a day of testing and learning I finally managed to bring the acceptance test and the facts to work. Please review again. Locally the acceptance tests are building fine with |
|
Can we merge this Change? |
vchepkov
left a comment
There was a problem hiding this comment.
I would pass a path to the config file instead of relying on user's home.
manifest uses jira::homedir, imho, that should be used
| EOS | ||
|
|
||
| pp = <<-EOS | ||
| # The output of `systemctl status postgresql` is non ascii which |
There was a problem hiding this comment.
I have some questions here. Why does that even break the exec resource? And is that a patch that should be added to the postgresql module instead?
There was a problem hiding this comment.
This I've copied from the existing default_parameters_spec.rb. I don't know if it is still needed.
7765047 to
b8e4777
Compare
OK - added that |
|
From my point of view, this MR can be merged: all tests are OK and all the requested changes are included. |
|
okay, nice, just was looking for this change 😃 will merge and release |
Fix breaking change regression of PR voxpupuli#429
Pull Request (PR) description
Optimization for PR Add "change database password" parameter - JIRA >10.3
This Pull Request (PR) fixes the following issues