V1.102.2
Important Updates
Kubecost recommends that any environment not using the Kubecost-provided Prometheus instance upgrade all clusters to Kubecost version 1.102+ due to a mismatched label that can cause container costs to be under-reported.
Additionally, Kubecost added a new recording rule in this version to enhance the CPU request-sizing recommendations accuracy when using the max algorithm. For clusters using existing Prometheus instances, this rule can be added by following this guide. Note that even without the recording rule, Kubecost will use a less efficient subquery in order to provide accurate recommendations.
Patch Release
- V1.102.2 fixes an issue where allocation data would return "Undistributable Idle" in scenarios where it should not have.