Replace use of CRD image defaults and configuration with global configuration (#1606)

* Replace the use of pkg/images.Image in internal handling, use cass-operators similar struct. Set up ImageRegistry to controllers and use the information from ImageRegistry to set the correct images.

* Modify perNodeConfig to use k8ssandra-client by default

* Remove unused code

* Add scripts to update Kustomize builds based on the changes from cass-operator

* Add CHANGELOG

* Add Helm charts changes (pointing to non-existant cass-operator chart version), go dependency updates

* Restrict ImageConfig loading to the operator namespace

* Update to cass-operator v1.27.0

* Remove flaky nil pointer causing assertion

* Make generate-imageconfig.sh work with MacOS sed also

* Replace cat with yq in the script for more consistent output, modify Helm charts indentation, update patch-image-config

* More Kustomize fixes, add missing component

* Fix medusa image coordinates

* Add new reconcile checkers (generation) for K8ssandraCluster and CassandraDatacenter to the test code, update to Kubernetes 1.33.5

* Fix unit tests, downgrade testify to 1.10.0 to prevent a deadlock

* Kubernetes 1.33.5 and Go 1.24.7

* Update to cass-operator v1.27.1

* Add Release Notes

Author: burmanm

CHANGELOG/CHANGELOG-1.27.md

@@ -15,6 +15,7 @@ When cutting a new release, update the `unreleased` heading to the tag being gen
 
 ## unreleased
 
+* [FEATURE] [#1605](https://github.com/k8ssandra/k8ssandra-operator/issues/1605) Container images, tags, repositories, registry and pullsecrets and now centrally managed in a ConfigMap with label `k8ssandra.io/config: image`. This is shared with the cass-operator and allows to configure everything from a single place. perNodeConfig is using k8ssandra-client as the image name.
 * [ENHANCEMENT] [#1591](https://github.com/k8ssandra/k8ssandra-operator/issues/1591) Remove the old medusa purge cronjob in favor of scheduled tasks
 * [ENHANCEMENT] [#1245](https://github.com/k8ssandra/k8ssandra-operator/issues/1245) Ensure ReplicatedSecret targets are cleaned up correctly
 * [BUGFIX] [#1603](https://github.com/k8ssandra/k8ssandra-operator/issues/1603) Fix the crd-upgrader to update cass-operator CRDs also as part of the Helm upgrade

CHANGELOG/RELEASE-NOTES.md

@@ -1,5 +1,26 @@
 # k8ssandra-operator - Release Notes
 
+## v1.27.0
+
+### Modifications to the configuration of images
+
+Starting from 1.27.0, the k8ssandra-operator will use the same ImageConfig structure as cass-operator. In the Helm charts, the configuration happens through the global.imageConfig property. It's divided to three sections, with the most important part usually being the `defaults` where we define the properties that are used by all the containers unless otherwise overridden:
+
+```
+defaults:
+  registry: "docker.io"
+  pullPolicy: IfNotPresent
+  # -- pullSecrets allow configuring the secret to use for pulling images from private registries.
+  # pullSecrets:
+  #   - my-secret-pull-registry
+```
+
+Changing any setting will apply to all images. For example, setting ``--set global.imageConfig.defaults.registry=privateregistry.local`` would pull all the images from `privateregistry.local` instead of `docker.io`. Setting a `pullSecret` would similarly allow pulling all images using that secret instead of having to define it separately for all container types.
+
+For more information, see the comments in the [Helm chart of cass-operator](https://github.com/k8ssandra/k8ssandra/blob/main/charts/cass-operator/values.yaml#L17).
+
+This also means deprecation of multiple fields in the CRD that were required to modify the image to be used such as `PerNodeConfigInitContainerImage`. To modify the image used by `perNodeConfigInitContainerImage`, modify the `k8ssandra-client` image. For Kustomize installations, the `imageConfig` is available in the `config/cass-operator/imageconfig` directory. 
+
 ## v1.15.0
 
 ### Deprecation of non-namespace-local MedusaConfigRef

Makefile

@@ -129,10 +129,13 @@ help: ## Display this help.
 
 ##@ Development
 
+CASS_OPERATOR_TAG ?= v1.27.1
+
 .PHONY: manifests
 manifests: controller-gen kustomize ## Generate WebhookConfiguration, ClusterRole and CustomResourceDefinition objects.
 	$(CONTROLLER_GEN) $(CRD_OPTIONS) rbac:roleName=k8ssandra-operator webhook paths="./..." output:crd:artifacts:config=config/crd/bases
 	KUSTOMIZE=$(KUSTOMIZE) ./scripts/prepare-helm-release.sh
+	KUSTOMIZE=$(KUSTOMIZE) CASS_OPERATOR_REF=$(CASS_OPERATOR_TAG) ./scripts/generate-imageconfig.sh
 
 .PHONY: generate
 generate: controller-gen ## Generate code containing DeepCopy, DeepCopyInto, and DeepCopyObject method implementations.

apis/k8ssandra/v1alpha1/k8ssandracluster_types.go

@@ -436,9 +436,8 @@ type DatacenterOptions struct {
 
 	// The image to use in each Cassandra pod for the (short-lived) init container that merges global and perNodeConfig
 	// configuration. This is only useful when PerNodeConfigMapRef is set.
-	// The default is "k8ssandra/k8ssandra-client:v0.8.1".
 	// +optional
-	// +kubebuilder:default="k8ssandra/k8ssandra-client:v0.8.1"
+	// DEPRECATED: use ImageConfig's k8ssandra-client instead
 	PerNodeConfigInitContainerImage string `json:"perNodeConfigInitContainerImage,omitempty"`
 
 	// The k8s service account to use for the Cassandra pods

apis/reaper/v1alpha1/reaper_types.go

@@ -85,10 +85,7 @@ type ReaperTemplate struct {
 	SecretsProvider string `json:"secretsProvider,omitempty"`
 
 	// The image to use for the Reaper pod main container.
-	// The default is "thelastpickle/cassandra-reaper:4.0.0-rc1".
 	// +optional
-	// TODO: update with real release version.
-	// +kubebuilder:default={repository:"thelastpickle",name:"cassandra-reaper",tag:"4.0.0-rc1"}
 	ContainerImage *images.Image `json:"containerImage,omitempty"`
 
 	// Deprecated: The main container image will be used for the init container as well.

charts/k8ssandra-operator/Chart.yaml

@@ -9,7 +9,7 @@ dependencies:
     version: 0.29.2
     repository: https://helm.k8ssandra.io
   - name: cass-operator
-    version: 0.60.0
+    version: 0.61.1
     repository: https://helm.k8ssandra.io
 home: https://github.com/k8ssandra/k8ssandra-operator
 sources:

charts/k8ssandra-operator/crds/k8ssandra-operator-crds.yaml

@@ -10004,11 +10004,10 @@ spec:
                               type: object
                           type: object
                         perNodeConfigInitContainerImage:
-                          default: k8ssandra/k8ssandra-client:v0.8.1
                           description: |-
                             The image to use in each Cassandra pod for the (short-lived) init container that merges global and perNodeConfig
                             configuration. This is only useful when PerNodeConfigMapRef is set.
-                            The default is "k8ssandra/k8ssandra-client:v0.8.1".
+                            DEPRECATED: use ImageConfig's k8ssandra-client instead
                           type: string
                         perNodeConfigMapRef:
                           description: |-
@@ -23080,11 +23079,10 @@ spec:
                         type: object
                     type: object
                   perNodeConfigInitContainerImage:
-                    default: k8ssandra/k8ssandra-client:v0.8.1
                     description: |-
                       The image to use in each Cassandra pod for the (short-lived) init container that merges global and perNodeConfig
                       configuration. This is only useful when PerNodeConfigMapRef is set.
-                      The default is "k8ssandra/k8ssandra-client:v0.8.1".
+                      DEPRECATED: use ImageConfig's k8ssandra-client instead
                     type: string
                   podPriorityClassName:
                     description: PodPriorityClassName defines the priority class name
@@ -29140,13 +29138,7 @@ spec:
                     type: object
                     x-kubernetes-map-type: atomic
                   containerImage:
-                    default:
-                      name: cassandra-reaper
-                      repository: thelastpickle
-                      tag: 4.0.0-rc1
-                    description: |-
-                      The image to use for the Reaper pod main container.
-                      The default is "thelastpickle/cassandra-reaper:4.0.0-rc1".
+                    description: The image to use for the Reaper pod main container.
                     properties:
                       name:
                         description: The image name to use.
@@ -35575,13 +35567,7 @@ spec:
                 - truststoreSecretRef
                 type: object
               containerImage:
-                default:
-                  name: cassandra-reaper
-                  repository: thelastpickle
-                  tag: 4.0.0-rc1
-                description: |-
-                  The image to use for the Reaper pod main container.
-                  The default is "thelastpickle/cassandra-reaper:4.0.0-rc1".
+                description: The image to use for the Reaper pod main container.
                 properties:
                   name:
                     description: The image name to use.

charts/k8ssandra-operator/values.yaml

@@ -12,7 +12,18 @@ global:
   # watchNamespaces: []
   # -- If true, the helm chart will generate ClusterRoles and ClusterRoleBindings.
   clusterScopedResources: true
-
+  # -- imageConfig modifications to the default settings provided by the cass-operator. One can override all the
+  # settings here. See https://github.com/k8ssandra/k8ssandra/blob/main/charts/cass-operator/values.yaml#L17 for more details.
+  imageConfig:
+    images:
+      reaper:
+        repository: "thelastpickle"
+        name: "cassandra-reaper"
+        tag: "4.0.0-rc1"
+      medusa:
+        repository: "k8ssandra"
+        name: "medusa"
+        tag: "0.25.0"
 # -- A name in place of the chart name which is used in the metadata.name of
 # objects created by this chart.
 nameOverride: ''
@@ -74,8 +85,8 @@ nodeSelector: {}
 ## Ref: https://kubernetes.io/docs/concepts/configuration/assign-pod-node/
 ##
 tolerations: []
-# -- k8ssandra-client provides CLI utilities, but also certain functions such as
-# upgradecrds that allow modifying the running instances
+# -- This client configuration is used for the CRD upgrader job only. Otherwise, modify
+# the global.imageConfig values to change k8ssandra-client
 client:
   # -- Uncomment to specify the name of the service account to use for the
   # client tools image. Defaults to <release-name>-crd-upgrader-k8ssandra.
@@ -86,13 +97,13 @@ client:
     # -- Image repository for the client
     repository: k8ssandra/k8ssandra-client
     # -- Tag of the client image to pull from
-    tag: v0.8.1
+    tag: v0.8.3
     # -- Pull policy for the client container
     pullPolicy: IfNotPresent
   # -- HTTPS proxy address to use for communication to helm.k8ssandra.io
   #proxy:
 # -- Allows managing CRD upgrades externally and fully disable the CRD upgrader job hook
 disableCrdUpgraderJob: false
+# cass-operator:
 # -- Allows disabling the check for the presence of cert-manager.
-cass-operator:
-  disableCertManagerCheck: false
+  # disableCertManagerCheck: false

config/cass-operator/cluster-scoped/kustomization.yaml

@@ -2,7 +2,10 @@ apiVersion: kustomize.config.k8s.io/v1beta1
 kind: Kustomization
 
 resources:
-- github.com/k8ssandra/cass-operator/config/deployments/cluster?ref=v1.26.0
+- github.com/k8ssandra/cass-operator/config/deployments/cluster?ref=v1.27.1
+
+components:
+  - ../imageconfig
 
 patchesStrategicMerge:
 - |-

config/cass-operator/imageconfig/fragments/medusa.yaml

@@ -0,0 +1,5 @@
+images:
+  medusa: 
+    repository: "k8ssandra"
+    name: "medusa"
+    tag: "0.25.0"