Given how much pain it was to setup the proper build job for base images, I'm wondering if we should move on with our CI/CD scripting to sth like dagger.io.
I tuned into an episode of the K8s podcast about it: https://kubernetespodcast.com/episode/236-dagger/
Eventually I had to write that shell script because Github Actions simply couldn't cut it. And I absolutely suffered from the effect of push and pray, as I also couldn't easily use act to testdrive the jobs locally.
And every now and then people are asking about how to run the usual CI jobs on their machines.
Also, we have a few rather complicated flows I set up that suffer from this push to cache, fetch from cache to pass around artifacts between builds.
And what happens if we move away from Github because Github is going down?
There is an experimental Java SDK they work on as well. So we could even try to write these directed acyclic graphs in our main language.
I really like that you can reuse other stuff from people that might be written in a different language.
I started listening. So far so good. I laughed at "push and pray" and how their community is basically "a support group for traumatized pipeline engineers". :rofl:
I like how they do a recap of the interview at the end.
Could Dagger be used to automate our release process?
That can be done in any CI "language".
The release process is, as far as I can tell, far away from being non-automateable
It's more of a political matter. People need to want to have it done that way.
Ok, @Steven Winship asked about automating it today. It deserves a dedicated topic, of course. I was just wondering if Dagger would be a good candidate.
If we want to try out Dagger, sure, why not.
If people feel reluctant to do such an important process in an experimental thing: just do it using GH actions YAML
I'd be happy with either one, honestly.
@Juan Pablo Tosca Villanueva this what I was telling you about. Dagger. ^^
Related, sort of: https://www.feldera.com/blog/the-pain-that-is-github-actions
Last updated: Nov 01 2025 at 14:11 UTC