Skip to content

Deploy a Pachyderm Cluster with CloudFront

After you have an EKS cluster or a Kubernetes cluster deployed with kops ready, you can integrate it with Amazon CloudFrontâ„¢.

Amazon CloudFront is a content delivery network (CDN) that streams data to your website, service, or application securely and with great performance. Pachyderm recommends that you set up Pachyderm with CloudFront for all production deployments.

To deploy Pachyderm cluster with CloudFront, complete the following steps:

  1. Create a CloudFront Distribution
  2. Deploy Pachyderm with an IAM role
  3. Apply the CloudFront Key Pair

Apply the CloudFront Key Pair

If you need to create signed URLs and signed cookies for the data that goes to Pachyderm, you need to configure your AWS account to use a valid CloudFront key pair. Only a root AWS account can generate these secure credentials. Therefore, you might need to request your IT department to create them for you.

For more information, see the Amazon documentation.

The CloudFront key pair includes the following attributes:

  • The private and public key. For this deployment, you only need the private key.
  • The key pair ID. Typically, the key pair ID is recorded in the filename.



The key-pair ID is APKAXXXXXXXXXXXXXXXX. The other file is the private key, which looks similar to the following text:



To apply this key pair to your CloudFront distribution, complete the following steps:

  1. Download the script from the Pachyderm repository:
curl -o
  1. Make the script executable:
chmod +x
  1. From the deploy.log file, obtain the S3 bucket name for your deployment and the CloudFront distribution ID.

  2. Apply the key pair to your CloudFront distribution:

./ --region us-west-2 --zone us-west-2c --bucket YYYY-pachyderm-store --cloudfront-distribution-id E1BEBVLIDYTLEV  --cloudfront-keypair-id APKAXXXXXXXXXXXX --cloudfront-private-key-file ~/Downloads/pk-APKAXXXXXXXXXXXX.pem
  1. Restart the pachd pod for the changes to take effect:
kubectl scale --replicas=0 deployment/pachd && kubectl scale --replicas=1 deployment/pachd && kubectl get pod
  1. Verify the setup by checking the pachd logs and confirming that Kubernetes uses the CloudFront credentials:
kubectl get pod

System Response:

NAME                        READY     STATUS             RESTARTS   AGE
etcd-0                   1/1       Running            0          19h
etcd-1                   1/1       Running            0          19h
etcd-2                   1/1       Running            0          19h
pachd-2796595787-9x0qf   1/1       Running            0          16h

kubectl logs pachd-2796595787-9x0qf | grep cloudfront
2017-06-09T22:56:27Z INFO  AWS deployed with cloudfront distribution at d3j9kenawdv8p0
2017-06-09T22:56:27Z INFO  Using cloudfront security credentials - keypair ID (APKAXXXXXXXXX) - to sign cloudfront URLs

Last update: April 5, 2021
Does this page need fixing? Edit me on GitHub