BillingS3MetricAbsent

This alert is used to be notified when S3 metrics used for billing are no longer present.

These are S3 bucket usage metrics calculated in the cloud zones and reported in the infra zones thanks to the federation.

Troubleshooting hints

This problem may have multiple possible causes :

• a problem with the Cloud Provider
• a problem with the exporter (in cloud zone)
• a problem on the federation between infra/consumption and the cloud zone
• a problem on the federation inside the cloud zone, between cloud/caascad and cloud/client subzones

You may check each of those possible causes :

• another alert may also raise if there is a federation problem between infra/consumption and the cloud zone. If there is no other alert, don't focus first on that federation.
• check if there is a federation problem between cloud/caascad and cloud/client subzones : check the metrics in Grafana Infra, in both Thanos and Thanos-client datasources. If both metrics exist, there is no federation problem.
  • the federation is defined in a servicemonitor in monitoring-client namespace.
  • check if it is correctly defined
  • check if Prometheus knows it in its configuration (kubectl -n monitoring-client port-forward svc/prometheus-operated 9090:9090 and on http://localhost:9090, check for the job monitoring-client/s3usage-federation/0.
• check the exporter in the cloud cluster : see if it is running, if it contains metrics. Also check its logs. The exporter is kubectl -n s3usage-exporter get pod -l app.kubernetes.io/name=s3usage-exporter