Conference:  KubeCon + CloudNativeCon North America 2022

Authors: Mayank Kumar, Andy Chen

2022-10-28

Many companies are thinking about automating application provisioning on Kubernetes but are stuck with old tools which cannot natively leverage the K8s apis and offer little scalability and extensibility. Mayank will show you how the Salesforce Hyper compute team that manages thousands of EKS clusters, is migrating Spinnaker pipelines that deploy K8s integrations like logging, monitoring and certificates to every EKS cluster on Hyperforce, to ArgoWorkflows. Mayank will talk through the existing problems and walk you through a step by step process, their team is using to convert all Spinnaker pipelines to Argo Workflow templates to improve reliability using custom retries, deployment velocity using memoization and using K8s features to gain more control over the various continuous deployment stages. He will also talk about a reusable library of ArgoWorkflow stages that is helping speed up the conversion process and making the developer experience around maintaining these deployment pipelines a true joy.

Conference:  KubeCon + CloudNativeCon North America 2021

Authors: Matteo Ruina, Guy Templeton

2021-10-14

The talk discusses the challenges of testing Kubernetes clusters and the potential solutions to build confidence in changes made to the clusters and their components.

• Operating Kubernetes clusters involves a large number of components, making it difficult to test the entire setup
• Running the full Kubernetes conformance test suite is slow and doesn't always cover the functionality users care about
• Skyscanner uses Kubernetes' existing test codebase along with Sonobuoy and Spinnaker to build a custom test suite
• The custom test suite allows cluster operators to focus on the functionality developers care about
• The talk provides an overview of potential solutions, including community-supported frameworks like Testemperor and Kubertest
• Kuberhealthy is a slightly different approach that uses an operator to run tests in-cluster and gather results
• Building confidence in changes made to Kubernetes clusters is crucial for reliability and frequent updates