November 2017 Summaries

The process of speeding up the creation of Kubernetes pods that require large Docker images can be enhanced by using a DaemonSet to pre-pull images, ensuring that the necessary Docker image is already on a node when a pod is scheduled to run. This approach is beneficial in scenarios where large images are involved and pods are frequently recreated, such as in scale-out scenarios. The method involves creating a Docker container that performs a pull command using a 'Docker in Docker' (dind) container, with the pod configured as a DaemonSet to run on every node. However, due to DaemonSet's limitation of always restarting pods, a workaround using an initContainer is employed to ensure the image is pulled only once, while a 'pause' container keeps the DaemonSet pod alive with minimal resource usage. This technique, although a workaround, is effective for pre-pulling Docker images to speed up Kubernetes pod deployment. The process is compatible with Codefresh, and a detailed implementation is available for reference.

The tutorial offers a step-by-step guide on configuring and deploying Kubernetes on AWS using the tool kops, providing a flexible alternative to managed solutions like Stackpoint Cloud. It begins with preparing the host environment by installing necessary software such as kubectl and kops, alongside configuring the AWS CLI for interaction with Amazon Web Services. The process includes setting up Route 53 domains for DNS management, configuring S3 buckets for storage, building the Kubernetes cluster configuration, and then deploying the cluster on AWS. Once deployed, the tutorial guides users on integrating the cluster with Codefresh for continuous deployment, emphasizing the ease of use and scalability of using kops with AWS for a production-grade Kubernetes cluster, despite it not being as comprehensive as fully-managed services like Google Kubernetes Engine (GKE).