Overview

LocalStack is a local AWS cloud environment that emulates core AWS services for development and testing.

When LocalStack is deployed on Kubernetes and configured to use the Kubernetes-native executor (by setting CONTAINER_RUNTIME=kubernetes), services that would normally spawn Docker containers (ie., Lambda, ECS, or RDS) instead create Kubernetes pods within the cluster. This enables dynamic scaling, isolation, and native Kubernetes orchestration.

Supported cases:

• Local Development Environments: Provide isolated, consistent environments for individual developers or small teams.
• Hosted Development Environments: Provide scalable and isolated development environments for teams.
• CI/CD Pipeline Testing: Run end-to-end integration tests in a reproducible, cloud-like environment during CI/CD workflows.

Requirements:

• K8s Cluster (such as k3d, minikube, EKS)
• kubectl
• (Optional) Helm
• (Optional) LS helm chart
• (Optional) LS operator

Deployment methods

LocalStack can be deployed into a Kubernetes cluster using multiple methods:

• using the LocalStack Operator
• using the LocalStack helm chart
• by manually creating Kubernetes manifests

The table below compares these methods.

Deployment approach	Pros	Cons
Operator	· Declarative, self-managed control plane · Built-in validation, defaulting, and reconciliation logic	· Requires in-cluster component (controller)
Helm chart	· Simplifies deployment using templates and values.yaml · Supports versioning, upgrades, and rollbacks · Supports both LocalStack Community and Pro images	· Customization is limited to chart values and overrides
DIY (YAML manifests)	· Full control over Kubernetes configuration and resources	· Time-consuming to set up and maintain · Manual updates and lifecycle management

Limitations

Service	Supported	Explanation
Sagemaker	No	Requires spawning Docker containers at runtime, which is unavailable when LocalStack runs inside a Kubernetes pod.
Bedrock	No	Requires spawning Docker containers at runtime, which is unavailable when LocalStack runs inside a Kubernetes pod.
EKS	No	Uses k3d, which requires Docker to create clusters. A Docker-free local mode is available but unsupported.
CodeBuild	No	Requires spawning Docker containers at runtime, which is unavailable when LocalStack runs inside a Kubernetes pod.
SSM	Partial	Most SSM functionality is supported. Session Manager (ssm:StartSession) for exec-ing into EC2 instances is not supported.
ECS	Partial	Core task execution is supported. FireLens log routing, volume mounts, port exposure from tasks, and private registry credentials via RepositoryCredentials are not supported.
EC2	Partial	Core EC2 functionality is supported. User data scripts may not behave identically to AWS — full parity is not guaranteed.
RDS	Partial	Most database engines are supported. MySQL 5.7 is not supported. Persistence across pod restarts is not supported, except for PostgreSQL and MariaDB.

Licensing

While the LocalStack Community image can be deployed in Kubernetes, it does not support containerized service functionality required for full cloud-like behavior. Services such as Lambda, which require spawning and managing compute containers at runtime, only support pod creation with the LocalStack Pro image. The Community edition lacks the runtime capabilities needed to dynamically instantiate and orchestrate workloads in Kubernetes.

Help & Support

LocalStack Enterprise (or additional purchase of the Kubernetes pack add-on) is the only version that provides:

• Official support and integration for Kubernetes environments.
• Dynamic pod creation by services like Lambda, ECS, RDS, etc.

Community image users:

• Can be deployed in Kubernetes cluster for basic service emulation (e.g., S3, SQS).
• Do not receive official Kubernetes support.
• Cannot use features like Lambda execution.