💪 Motivation

We don't currently support any particular logging paradigm - pipeline nodes simply forward stdout and stderr from users' application code to the pod's respective output stream. This makes logs accessible in the following manner:

$ kubectl logs [-n <namespace>] <pod-id> [-f]

This is necessary for all sorts of logging frameworks, such as those that use sidecar containers to tail pod logs & forward them to an external service such as logstash or cloudwatch, but it doesn't make our solution complete. There might be other ways to make log collection & insights easier, such as providing an optional UI + ELK stack, or at least providing some examples of how logging might be managed in both development and production environments.