How to Develop a Robust Operator for Day-2 Operations - Lessons from KubeVirt and HCO

Explore lessons learned from developing the Hyperconverged Cluster Operator for KubeVirt in this 35-minute conference talk. Delve into the complexities of creating a robust operator for day-2 operations, focusing on managing stateful applications and virtual machines. Learn about control plane vs. workload upgrades, long-running upgrades, and reliability concerns such as canary deployments and fail-forward upgrades. Discover strategies for protecting pre-release features, introducing new APIs, deprecating old ones, and discriminating between defaults and user choices. Gain insights on implementing a declarative approach, minimizing the upgrade matrix, and plotting the upgrade graph. Equip yourself with the knowledge to develop trustworthy operators capable of handling fully automatic and continuous upgrades in complex environments.