A Kubernetes pod may request 25× more CPU than you think
Dynatrace describes an example where a dashboard shows 40 millicores requested for an api-server container, while Kubernetes’ scheduler reserved 1,000 millicores—a full CPU core.
TLDR
The New Stack shared a Dynatrace article warning that a Kubernetes pod may request 25× more CPU than you think. Dynatrace illustrates the discrepancy with an api-server container: its dashboard shows a 40-millicore request, but the scheduler reserved 1,000 millicores, or one CPU core.
Combined views
519
1 Source, first seen 7h ago
saves