From my experience, it seems that most people throw an application onto Kubernetes (either with Helm or manually) and then think they can call it a day. Through our use of Kubernetes at GumGum, we have encountered a number of “gotchas” that we wanted to list here to help you cover your bases before launching your app on Kubernetes.
We will start with configuring a clean environment in which our pods can run. Kubernetes does a fantastic job at handling pod scheduling and failure states. One thing we learned, however, is that the Kubernetes scheduler can sometimes have a hard time placing pods if it can’t gauge how many resources that pod needs to run successfully. This is where resource requests and limits come in. There is much debate on the best approach to setting app requests and limits; it really can feel like more of an art than a science. …
About