You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This got me thinking dagger could help with some pain points I have with the current pipeline
Pipeline logic can't easily be run locally
PRs from forked repos fail because they can't access the kubeconfig secret (for very sensible security reasons :)
I did some hacking on this over the holidays and got our pipeline running in a k3s container managed by the module.
Benefits
Whole pipeline including deploying falco and creating benchmark jobs can be run locally in a kind cluster
Whole pipeline can be developed in go. GitHub Actions workflow just becomes a wrapper that calls the dagger pipeline
Automated tests can be run with full stack (including kepler in estimation mode) in a k3s container on a GitHub Actions runner
k3s test clusters use same flux manifests as prod cluster (just needs patches to helmrelease values for volume mounts)
Weaknesses
There is a learning curve for dagger
Barrier to entry for new contributors may be slightly higher (having go and docker knowledge helps)
Alternatives considered
act is another tool for running actions workflows locally but using it with kind needs a "kitchen sink" docker image that is 18Gb! nektos/act#335
Due to size I've split the changes into 2 PRs. #152 ports the current pipeline to dagger. #153 adds the integration test and docs for running the pipeline locally.
I did invest some time here because I think its a good solution for us. However it was also fun to learn something new and I'll probably use dagger for some of my own projects. So if we don't go this direction no worries! ☺️
The text was updated successfully, but these errors were encountered:
Before the break I got a tip from @dipankardas011 about this newish k3s module for dagger
https://daggerverse.dev/mod/github.com/marcosnils/daggerverse/k3s
This got me thinking dagger could help with some pain points I have with the current pipeline
I did some hacking on this over the holidays and got our pipeline running in a k3s container managed by the module.
Benefits
Weaknesses
Alternatives considered
act
is another tool for running actions workflows locally but using it with kind needs a "kitchen sink" docker image that is 18Gb!nektos/act#335
Due to size I've split the changes into 2 PRs. #152 ports the current pipeline to dagger. #153 adds the integration test and docs for running the pipeline locally.
I did invest some time here because I think its a good solution for us. However it was also fun to learn something new and I'll probably use dagger for some of my own projects. So if we don't go this direction no worries!☺️
The text was updated successfully, but these errors were encountered: