Hands-On Microservices with Kubernetes

Centralized logging with Kubernetes

In Kubernetes, containers write to the standard output and standard error streams. Kubernetes makes those logs available (for example, via kubectl logs). You can even get logs of the previous run of a container if it crashed by using kubectl logs -p, but, if the pod is rescheduled, then its containers and their logs disappear. If the node itself crashes, you'll lose the logs too. Even when all the logs are available for a cluster with a lot of services, it is a non-trivial task to sift through the container logs and try to make sense of the state of the system. Enter centralized logging. The idea is to have a log agent running, either as a side container in each pod, or as daemon set on each node, listen ...