Running Kubernetes in production has moved from theory to practice in many organizations. Day 2 operations, however, are still challenging and mostly involve custom approaches. Finding an off-the-shelf solution is complicated by the fact that most logging products do not support Kubernetes-native concepts. This is why we felt the need to create the open source Logging Operator project. It is 2 years old and is deployed in a number of production environments both by us and the community. Some of the key features are name space isolation, pod label selector support, and using secrets for credentials. Sandor will introduce the challenges of configuring a complex logging stack on Kubernetes and will dive into the lessons learned during the creation and deployment of the operator.