Kubernetes out-of-memory (OOMKilled) issues can be challenging to troubleshoot, as demonstrated by Robert Wunderer, founder of CapriSys GmbH, who encountered such a problem after deploying a new software version in a multi-tenant e-commerce system. Despite having adequate resource settings in place, the rollout led to high HTTP failure rates across tenants due to the report service pods being killed for memory shortages. Using Dynatrace, Robert identified an increased memory requirement in the new version, which was causing the pods to hit their memory limits and restart. By increasing the memory limit from 400MB to 600MB, the issue was resolved, highlighting the need for proper resource allocation and testing in Kubernetes environments. The case underscores the importance of observability tools like Dynatrace in quickly identifying root causes and ensuring smooth operations, while also emphasizing the value of comprehensive testing to prevent future issues.