Kubernetes containers with short run times not being reported

Products

• CloudHealth

Issue/Introduction

Any run times shorter than 15 minutes may not be reported.

Resolution

The CloudHealth Kubernetes agent takes a snapshot every 15 mins via the API.  The API lists all nodes/pods which are currently running.  Once the container has been stopped, it typically is not returned in this API call.  If you have very short-running pods (~1 min in runtime), the platform would typically miss collecting data for this pod unless the agent happened to run at the same time the pod was running.