The 1.73.0 release has heavy focus on performance improvements, along with a number of new features.
- Fix performance regression on overview and cost allocation page for queries 2d and higher
- General performance improvements for larger clusters: fewer Prometheus queries and more use of the ETL cache
- Kubecost health alerts: Kubecost will now send a slack alert if it can't reach prometheus or emit metrics. You can enable this by setting a slack webhook here and setting notifications.alertConfigs.kubecostHealth
- Reconciled AWS cloud asset pricing now flows to idle allocations
- Fix for high network cost estimates
- Support querying by annotation in the API
- Credits now reflected on GCP out-of-cluster assets page
- Grafana no longer runs as admin user by default (requires a manual restart of grafana to pick up changes)
- Support supplying NameIDFormat in SAML from helm chart
- Kubecost no longer runs anything as root by default, and consistently uses securityContext group 1001
- Default ETL cache time period to 90d instead of 120d