operator

README

Calico Operator

This repository contains a Kubernetes operator which manages the lifecycle of a Calico or Calico Enterprise installation on Kubernetes or OpenShift. Its goal is to make installation, upgrades, and ongoing lifecycle management of Calico and Calico Enterprise as simple and reliable as possible.

This operator is built using the operator-sdk, so you should be familiar with how that works before getting started.

Get Started Developing

Code structure

There are a few important areas to be aware of:

• Operator API definitions exist in api/v1
• Rendering code for generating Kubernetes resources is in pkg/render
• Control/reconcile loops for each component can be found in pkg/controller/<component> There is a layer that was introduced with the upgrade to operator-sdk v1.x of controllers in controller that currently calls pkg/controller/<component>.
• Status reporting is in pkg/controller/status

Tests:

• Tests for file X.go can be found in X_test.go.
• FV tests which run against a local cluster can be found in test/*.go.

Controller Dependency Graph

This graph shows the dependencies between controllers. Optional dependencies are in dashed lines.

Design principles

When developing in the operator, there are a few design principles to be aware of.

• API changes should be rare occurences, and the API should contain as little as possible. Use auto-detection or automation wherever possible to reduce the API surface.
• Each "component" should receive its own CRD, namespace, controller, and status manager. e.g., compliance, networking, apiserver.
• Controllers interact with each other through the Kubernetes API. For example, by updating status on relevant objects.

Adding a new CRD

New APIs are added using the operator-sdk tool.

operator-sdk create api --group=operator --version=v1 --kind=<Kind> --resource

When modifying or adding CRDs, you will need to run make gen-files to update the auto-generated files. The tool might change the scope of existing resources to "Namespaced", so make sure to set them back to their desired state.

Adding a new controller

New controllers are also added using the operator-sdk tool.

operator-sdk create api --group=operator --version=v1 --kind=<Kind> --controller

New controllers will be created in the newer format so it should be considered if it is desirable to keep the current format that calls to a controller in pkg/controller or add the controller only in controllers.

Running it locally

You can create a local k3d cluster with the Makefile:

make cluster-create

Then, run the operator against the local cluster:

# enable-leader-election is necessary since you'll be running the operator outside of a cluster
KUBECONFIG=./kubeconfig.yaml go run ./ --enable-leader-election=false

To launch Calico, install the default custom resource:

kubectl create -f ./config/samples/operator_v1_installation.yaml

To tear down the cluster:

make cluster-destroy

Using Calico Enterprise

To install Calico Enterprise instead of Calico, you need to install an image pull secret, as well as modify the Installation CR.

Create the pull secret in the tigera-operator namespace:

kubectl create secret -n tigera-operator generic tigera-pull-secret \
    --from-file=.dockerconfigjson=<PATH/TO/PULL/SECRET> \
    --type=kubernetes.io/dockerconfigjson

Then, modify the installation CR (e.g., with kubectl edit installations) to include the following:

spec:
  variant: TigeraSecureEnterprise
  imagePullSecrets:
  - name: tigera-pull-secret

You can then install additional Calico Enterprise components by creating their CRs from within the ./deploy/crds/ directory.

Running unit tests

To run all the unit tests, run:

make test

To run a specific test or set of tests, use the GINKGO_FOCUS argument.

make test GINKGO_FOCUS="component function tests"

Making temporary changes to components the operator manages

The operator creates and manages resources and will reconcile them to be in the desired state. Due to the reconciliation it does, if a user makes direct changes to a resource the operator will revert those changes. To enable the user to make temporary changes, an annotation can be added to any resource directly managed by the operator which will cause the operator to no longer update the resource. Adding the following as an annotation to any resource will prevent the operator from making any future updates to the annotated resource:

Do not use this unless you are a developer working on the operator. If you add this annotation, you must remove it before the operator can manage the resource again.

unsupported.operator.tigera.io/ignore: "true"

Example update to calico-node DaemonSet

Notice that the annotation is added in the top level metadata (not in the spec.template.metadata). (note the below is not a valid manifest but just an example)

kind: DaemonSet
apiVersion: apps/v1
metadata:
  name: calico-node
  namespace: calico-system
  labels:
    k8s-app: calico-node
  annotations:
    # You should NOT use this unless you want to block the operator from doing its job managing this resource.
    unsupported.operator.tigera.io/ignore: "true"
spec:
  template:
    metadata:
      labels:
        k8s-app: calico-node
      annotations:
        scheduler.alpha.kubernetes.io/critical-pod: ''
    spec:
      containers:
        - name: calico-node
          image: calico/node:my-special-tag