This directory contains examples of different kinds of kubeapply configs. The
sections below describe how to run through a typical kubeapply workflow and
provision the associated resources in a small, local cluster.
Make sure you've installed kubeapply and all of the dependencies described in the
main README.
Then, install kind, which we'll use to run a
local cluster. It's also possible to run this example against something like
minikube or a real, remote cluster,
but you'll need to alter some of the steps below.
The following are included at the top-level:
| Path | Description |
|---|---|
cluster-local1.yaml |
Cluster config for first version of our test cluster. Contains basic information about the cluster as well as parameters for template expansion and/or starlark evaluation. |
cluster-local2.yaml |
Cluster config for alternative version of the test cluster. Uses same profile as first version, but has different expansion parameters. Not used in sample workflow below, but left as an example of how different cluster instances in different environments and/or regions might be configured. |
charts |
Contains an example envoy chart that's referred to in the profile. kubeapply can also source charts from git, http, or s3. |
expanded |
Expanded Kubernetes configs generated by running kubeapply expand. These can be applied in the cluster by running kubeapply apply or directly via kubectl apply. |
profile |
Source templates and starlark files for our cluster, used by kubeapply expand to fill out the expanded directory. |
The profile directory contains configs for three "services", each of which demonstrates a
different way of specifying configs for expansion:
| Service | Description |
|---|---|
echoserver |
Uses plain Kubernetes YAML and go-templated YAML. |
envoy |
Uses the envoy Helm chart to generate configs for an Envoy deployment. |
redis |
Uses skycfg, which evaluates starlark to generate configs. |
The steps below use the provided Makefile for
convenience, but you can also just run the associated kubeapply commands directly.
Starts up a local cluster via kind and dumps a kubeconfig in .kube/kubeapply-test-cluster.yaml.
Expands configs from profile into the expanded directory for cluster-local1.
It should be a no-op from the git perspective unless you've made changes in the cluster
config or profile files.
Runs kubeconform over the expanded configs to validate that they are legitimate Kubernetes
configs before continuing.
Runs kubectl diff against the local cluster and shows the results in a pretty format in the
terminal.
Runs kubectl apply against the local cluster and outputs a table that summarizes the changes
that were made.
Run kubectl --kubeconfig=.kube/kubeapply-test-cluster.yaml get pods --all-namespaces, etc.
to verify that the apply worked correctly!
Stops local cluster created in first step and deletes the kubeconfig.