Yesterday, I was running into an issue while trying to register a new GitLab CI Runner with our self-hosted GitLab instance. While that worked fine about 2 weeks ago, it did not this time. The only difference is, that we recently upgraded to version 15.9.2 of GitLab.
There are cases where I do not want to clone a complete project just to make a small change. GitHub and GitLab both provide a browser-based IDE.
Earlier this year we decided to test Renovate to manage automated dependency updates in our self-hosted GitLab environment. Being able to run Renovate in our own environment and configure it to our needs made a lot of sense. Since we run our internal tooling on a Nomad cluster, we had to configure a Nomad job for Renovate.
Testing GitLab CI build pipelines can be a bit annoying. You must make your changes, commit and push them to kick off the CI pipeline. Then you have to wait a while for the result to show up and start over again if something went wrong. If your CI runners are busy, you keep waiting and waiting for the next free slot. Luckily, it is possible to run GitLab CI jobs completely locally. After installing the gitlab-runner package locally, you can execute a job like this:
When we made the move to GitLab 1,5 years ago, it was clear to me that we would need some automation to simplify the creation of groups and projects and to sync the LDAP group memberships to the matching GitLab groups. I did a quick search on Packagist for GitLab client libraries and found the m4tthumphrey/php-gitlab-api package.
Last year GitLab introduced the Review Apps feature. Review Apps are app environments that are created dynamically every time you push a new branch up to GitLab. As a bonus point the app environments are automatically deleted when the branch is deleted. Since we moved to using docker for quite a few of our projects I was keen on figuring out how to combine Docker and the GitLab Review Apps functionality as the documentation only mentions NGINX as a way to run Review Apps. As it turns out, it is rather simple to deploy docker containers as a Review App.
In our "old" Jenkins set-up things were simple: The Jenkins master and Satis were running on the same host thus Jenkins could easily invoke Satis via a command-line call. Unfortunately GitLab does not allow that. The only option which is currently available in GitLab is to trigger Satis via a webhook. Neither Satis itself or Satisfy which we actually use provide support for webhooks. Thus we extended Satisfy with a simple controller which invokes the Satis cli command. Definitely not the best solution but it works for us.
In my recent attempt to migrate away from our Jenkins infrastructure to the new GitLab CI Runner infrastructure I ran into a problem: Since we want to use Docker images for the GitLab CI builds I struggled a bit on how pass the authentication information for Satis and GitLab into the docker images. Since the base images - basic PHP setup - should be used for our projects I did not want to share the access credentials in the different base images. Gitlab's secret variables sounded like a good idea but unfortunately they need to be defined for each and every project. Currently we have more than 250 projects in our GitLab instance, configuring secret variables for all the projects would have been a big pain.
We recently started experimenting with the GitLab CI Runners as we are looking to replace our "old" Jenkins v1.x set up with something new. And since over the last few weeks we had some issues with the Jenkins GitLab plugin we thought it might a good idea to take a deeper look into the GitLab CI Runners. One the plus side the GitLab CI Runners are configured via a YAML file in a similar fashion as you would configure Travis CI which we use to build our open-source components. Since on Travis we rely a lot on the so-called matrix builds to run the unit tests for different PHP versions I was wondering how we could solve the problem with GitLab CI. At first glance GitLab CI does not have a matrix build command but it comes with a feature called Anchors which kind of act as a template that can be merged in a job configuration.