In a containerized setup we have to think of the following areas for logging (starting at the bottom):
- host-level and intra-host (networking), that is, system logs and service logs on the host machine as well as logs that capture communication between different hosts.
- container-level and intra-container (networking), that is, how are your containers doing, depending on the type of runtime this might be on the, say Docker engine level or some other supervisor process.
- service or app-level as well as business-related stuff, that is, the logs you generate in the service running in the container and its communication aspects. In this context, both service-level output (for example, a log entry that captures a certain update of a datastore) but also business relevant stuff (e.g. the transaction amount) are of interest and not necessarily always the same.
For starters, to get a nice overview and introduction to the topic, you can read the following articles: