Deploying Workloads: Checklist

Objective

Provide a checklist of commonly missed steps to follow whenever creating a new workload on the Panfactum Stack.

Checklist

• Have you ensured that the container image is sourced from ECR (either directly or via the pull-through cache)?

• Have you implemented liveness and readiness probes?

• Have you ensured your workload can gracefully terminate in under 90 seconds?

• Have you set the appropriate priority class?

• Have you set resource minimums for your workload?

• Have you enabled the appropriate node classes?

• Have you ensured you protected secret configuration values by providing them via common_secrets or secret_mounts (vs. common_env or config_map_mounts)?

• Have you created the necessary temporary directories so that you can leave the root filesystem as read-only?

• Have you verified that the container can run under uid 1000? If it cannot, have you set the uid input?

• Have you given the workload all the necessary permissions?

• Have you chosen an SLA level for your workload and completed the necessary checklist?

• Do you have a networked workload? If so:

  • Have you set the ports for each container?

  • Do you need to allow public access? Have you set up an Ingress with kube_ingress?