Connect your AWS ParallelCluster to Clusterra for unified HPC management.
- Python 3.10+
- AWS CLI configured (
aws configure) - OpenTofu (
tofu) or Terraform - AWS ParallelCluster (
pcluster) installed via pip
The easiest way to connect is using our interactive installer, which handles infrastructure deployment and API registration automatically.
# 1. Clone the repository
git clone https://github.com/clusterra/clusterra-connect.git
cd clusterra-connect
# 2. Install dependencies
pip install -r requirements.txt
# 3. Run the installer
python3 install.py- Configuration: Interactive prompts to select your Region, VPC, and Subnets (supports both new cluster creation and existing clusters).
- Infrastructure: Uses OpenTofu to deploy a VPC Lattice service association, enabling secure communication between your cluster and Clusterra.
- Registration: Automatically registers your cluster with the Clusterra API using your Tenant ID.
If you already have a running ParallelCluster:
- Select "Existing Cluster" in the installer.
- Provide the Head Node Instance ID (e.g.,
i-0123456789abcdef0). - The installer will inspect the instance to identify the correct VPC and Subnet, then deploy the necessary connectivity layer.
Clusterra Connect uses AWS VPC Lattice for secure, private connectivity without exposing your cluster to the public internet.
- Private: No public IPs required for API communication.
- Secure: Cross-account access is strictly controlled via IAM and Lattice Service Network policies.
- Automated: The installer handles the Lattice Service association and RAM resource share acceptance.
Ensure your AWS CLI is configured for the correct account and region:
aws sts get-caller-identityIf the installer fails during the infrastructure phase, you can inspect the logs or try running Tofu manually for more details:
tofu init
tofu apply -var-file=generated/terraform.tfvarsThe installer waits for the RAM Resource Share invitation from Clusterra. If it times out, you can check for pending invitations in the AWS RAM Console.
To remove all resources and disconnect the cluster:
# Destroy cloud resources
tofu destroy -var-file=generated/terraform.tfvars
# If you created a new cluster via the installer, also delete it:
pcluster delete-cluster --cluster-name <your-cluster-name> --region <region>