When looking at deployments, we have pre-deployment strategies to mitigate the risk of breaking things, like tests and progressive delivery. Regarding the post-deployment phase, we are not as strong with methods and techniques as in the pre-deployment phase.

Progressive delivery already touches the post-deployment phase. But there are some more measurements we can consider to create a post-deployment strategy. Dashboards with metrics logs and traces are one-way but not stateful. The presentation will present a concept using Argo Rollouts for progressive delivery and leverage Argo CD notifications after a deployment to collect telemetry data with OpenTelemetry and Prometheus statefully. It presents a concept around function-level metrics based on the codebase that helps collaborate better with the application engineers if things go wrong after a deployment. In the post-deployment phase, we can’t mitigate the risk of breaking things, but we can reduce the effort of solving them.