fix: kube-rbac-proxy defunct registry fix and dynamic-link library fix

[richscott]

Description

These changes fix two issues that cause the creation failure of a local cluster by the kind-all Make target.

1. The gcr.io registry no longer hosts the kube-rbac-proxy:v0.16.0 image (that registry appears to have other, much older versions but not this one, nor any newer/higher version) - the two mentions of this image have been replaced by using docker.io instead.
2. After that is fixed, there is a subsequent error where the armada-operator-controller pod immediately fails, with a message of /armada-operator: not found. After some research and testing, I verified that /armada-operator was indeed there, had correct permissions, etc. Replacing the scratch image base with alpine, then manually running a shell to attempt to run /armada-operator also failed with the same message.

Some more online research yielded a possible solution of changing the Dockerfile for the image with:

diff --git a/Dockerfile b/Dockerfile
index 0f8d47a..d61bc20 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -1,4 +1,5 @@
-FROM scratch
+FROM alpine:latest
+RUN apk add libc6-compat

 COPY armada-operator /

This is due to the MUSL libc implementation being incompatible with the GNU libc implementation that most full-sized Linux systems have (e.g. Ubuntu). This did fix the problem, however it seemed like more of a change than would be preferred. I then reverted to using the scratch image, and instead changed the compilation to make a statically-linked binary:

 .PHONY: build
build: generate ## Build armada operator binary.
-       go build -o $(LOCALBIN_APP)/armada-operator cmd/main.go
+       CGO_ENABLED=0 go build -o $(LOCALBIN_APP)/armada-operator cmd/main.go

.PHONY: build-linux-amd64
build-linux-amd64: generate ## Build armada operator binary for linux/amd64.
-       GOOS=linux GOARCH=amd64 go build -o $(LOCALBIN_APP)/armada-operator cmd/main.go
+       CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -o $(LOCALBIN_APP)/armada-operator cmd/main.go

This then worked correctly. All unit and integration tests passed, and make build docker-build kind-all-dev successfully builds and starts a functioning Armada cluster again. There are no user-visible changes with this PR.

Fixes #378

Type of change

Please select the type of change your PR introduces:

• Bugfix
• Feature
• Code Style Update (formatting, renaming)
• Refactor (code changes that do not fix a bug or add a feature)
• Documentation Update
• Other (please describe):

How Has This Been Tested?

make test-unit, make test-integration, make build docker-build kind-all-dev all succeeded.

• Test Configuration:

  • Kubernetes Version: kind v0.31.0 go1.25.8 linux/amd64
  • Helm Version: version.BuildInfo{Version:"v3.18.4", GitCommit:"d80839cf37d860c8aa9a0503fe463278f26cd5e2", GitTreeState:"clean", GoVersion:"go1.24.4"}
  • OS: Ubuntu 24.04.4 LTS

• Test Steps:

  1. make test-unit
  2. make test-integration
  3. make build docker-build kind-all-dev, then verify all the expected Armada K8s pods are running, e.g. kubectl get pods -A.

Checklist:

• My code follows the style guidelines of this project.
• I have performed a self-review of my code.
• I have commented on my code, particularly in hard-to-understand areas.
• I have made corresponding changes to the documentation.
• My changes generate no new warnings.
• I have added tests that prove my fix is effective or that my feature works.
• New and existing unit tests pass locally with my changes.
• Any dependent changes have been merged and published in downstream modules.