Progressive Delivery

[jeffmccune]

Write an in-depth topic describing how Kargo pairs nicely with Holos and publish it on the docs site.

Whoever picks this up next, the setup notes are at:

https://github.com/holos-run/bank-of-holos/blob/main/docs/kargo.md

Then, holos render platform -t $USER in the bank of holos repo should render the kargo manifests for the service and promotion steps.

The only thing being promoted is the bank front end.

See https://news.ycombinator.com/item?id=42382850 for an explanation I gave of the lay of the land.

[jeffmccune]

This is wired up and working in 7b6a0c7 of https://github.com/holos-run/kargo-demo

Next steps are to write up the documentation and then record a video about it.

[jeffmccune]

The reusable definition is #KargoProjectBuilder and an example of using it is at https://github.com/holos-run/kargo-demo/blob/v0.0.1/projects.podinfo.cue#L5

[jeffmccune]

Todo:

• dev should be dev-podinfo at https://github.com/holos-run/kargo-demo/blob/v0.0.1/deploy/projects/podinfo/components/kargo-project/kargo-project.gen.yaml#L10
• branch should be scoped to the project and maybe also the component within the project at https://github.com/holos-run/kargo-demo/blob/v0.0.1/deploy/projects/podinfo/components/kargo-stages/kargo-stages.gen.yaml#L16
• Might as well put the manifest at the root instead of deep in the tree at https://github.com/holos-run/kargo-demo/blob/v0.0.1/deploy/projects/podinfo/components/kargo-stages/kargo-stages.gen.yaml#L30
• Move the requestedFreight definition from here to the stages field just above on Line 10.
• Move the requestedFreight configuration from here to stages here

[jeffmccune]

Merged and released unifying yaml files with the configuration from #376 and #377 in Holos v0.101.7 so we can integrate nicely with Kargo's yaml-update step.

See also the discussion I had with Dan from CUE regarding this in the Unify YAML Files thread.

[jeffmccune]

The simple case of promoting a container image works well for podinfo and httpbin.

Exploring the use case of promoting cert-manager when new helm charts are released.

These are the two main use cases I want to document. Watching for updated helm charts and automatically producing a pull request with the fully rendered diff, and promoting a container image.

[jeffmccune]

cert-manager and istio both update nicely with the github action

Full reset takes about a minute.

time bash -c './scripts/reset-cluster && ./scripts/apply'

Trigger the process by deleting the stage and adding it back.

kubectl delete -n istio stages.kargo.akuity.io main
kubectl apply --server-side=true -f deploy/components/istio-promoter/istio-promoter.gen.yaml