datadog-agent/doc/tutorials/dev/lab/kind.md
Quentin de Longraye 141d11575b
refactor(e2e): replace GCR image references with official registries (#60424)
### What does this PR do?

Replaces explicit GCR image-pull references in the E2E framework and tests with official registry endpoints:

- Datadog public image defaults and SSI workloads use `registry.datadoghq.com`.
- Istio ambient images use the existing Docker Hub cache via `InternalDockerhubMirror()`, with pull credentials for `istiod`, CNI, `ztunnel`, and generated waypoints.
- Documentation, image-resolution expectations, and standard-Agent repository recognition follow those endpoints.

Existing image version tags, account-level AWS cache configuration, IAM permissions, and Helm chart URLs are unchanged.

### Motivation

Move image consumers away from GCR and reuse the existing Docker Hub cache for Istio, following the investigation in #58215.

### Describe how you validated your changes

- Agent image-resolution unit tests passed through `dda inv test`.
- Compiled the affected environment packages, Istio component, SSI tests, and NPM suite with Bazel's `compilation_outputs` output group.
- Verified that SSI workload tag `d425e7df` has the same digest on GCR and the official Datadog registry; verified availability of Istio `pilot:1.30.8` and the documentation's Agent image tag at their replacement endpoints.
- Rendered the pinned Istio 1.30.8 charts and verified cached images and pull secrets for `istiod`, CNI, and `ztunnel`, plus the registry and credentials inherited by the generated waypoint template.
- Internet-access registry linter, Python syntax check, and `git diff --check` passed.


Co-authored-by: quentin.delongraye <quentin.delongraye@datadoghq.com>
2026-11-16 12:01:04 +00:00

2.8 KiB

Kind


Kind (Kubernetes in Docker) creates lightweight local Kubernetes clusters for development and testing.

Prerequisites

Quick start

# Create a local Kind lab environment (installs the Agent by default)
dda lab local kind --id dev

Create cluster

# Create a cluster without installing the Agent
dda lab local kind --id dev --no-agent

# With specific Kubernetes version
dda lab local kind --id dev --k8s-version v1.30.0

# Recreate existing cluster
dda lab local kind --id dev --force

Deploy agent

By default, dda lab local kind installs the Datadog Agent via Helm. Provide an API key with E2E_API_KEY or ~/.test_infra_config.yaml (see Lab environments).

From registry

# Default agent image (Agent installation is enabled by default)
dda lab local kind --id dev

# Custom image
dda lab local kind --id dev --agent-image registry.datadoghq.com/agent:7.50.0

# With custom Helm values
dda lab local kind --id dev --helm-values ./values.yaml

Build locally (custom build command)

If you want to build a local Agent image using a custom command, pass --build-command.

/// note The build runs inside a developer environment (see Using developer environments). Ensure it is started first:

dda env dev start

///

# Example: build an image tagged datadog/agent-dev:local, then load+install it
dda lab local kind --id dev \
  --build-command "dda inv agent.hacky-dev-image-build --target-image datadog/agent-dev:local"

Load existing image

# Load pre-built local image
dda lab local kind --id dev --load-image myagent:dev

Working with the cluster

# Switch kubectl context
kubectl config use-context kind-<id>

# Check agent pods
kubectl get pods -n datadog

# View agent status
kubectl exec -n datadog daemonset/datadog-agent -- agent status

# View agent logs
kubectl logs -n datadog -l app=datadog -f

Options

Option Description
--id, -i Environment id
--k8s-version Kubernetes version (default: v1.32.0)
--no-agent Do not install the Datadog Agent
--agent-image Custom agent image
--load-image Load existing local docker image into the cluster
--helm-values Path to custom Helm values.yaml file
--build-command Command to build the agent image (must output an image tagged datadog/agent-dev:local)
--devenv Developer environment ID (see dda env dev)
--force, -f Recreate cluster if exists
--nodes-count Number of nodes in the cluster (default: 2)