#Monitoring Ceph RGW with the Ceph RGW Exporter

#Install the Operator

1. Go to Administrator.

2. In the left navigation bar, click Marketplace > OperatorHub.

3. Find Alauda Storage Exporter and click Install. Select the alpha channel, install it into the rook-ceph namespace (the namespace where Rook-Ceph runs), and set the upgrade approval to Manual. For a detailed description of the installation dialog, refer to Operator.

   Alternatively, apply a Subscription from the command line on the control plane node of the cluster. The rook-ceph namespace already contains an OperatorGroup created by Rook-Ceph, so no new one is needed.

   cat << EOF | kubectl apply -f -
   apiVersion: operators.coreos.com/v1alpha1
   kind: Subscription
   metadata:
     name: ceph-rgw-exporter
     namespace: rook-ceph
   spec:
     channel: alpha
     installPlanApproval: Manual
     name: ceph-rgw-exporter
     source: platform
     sourceNamespace: cpaas-system
   EOF

   Approve the generated InstallPlan:

   ip=$(kubectl -n rook-ceph get subscription ceph-rgw-exporter -o jsonpath='{.status.installPlanRef.name}')
   kubectl -n rook-ceph patch installplan $ip --type=merge -p '{"spec":{"approved":true}}'

4. Verify that the Operator is running.

   kubectl -n rook-ceph get csv | grep ceph-rgw-exporter
   kubectl -n rook-ceph get deploy ceph-rgw-exporter-operator

   Expected output:

   ceph-rgw-exporter.v1.0.3   Alauda Build of Ceph RGW Exporter   1.0.3   Succeeded
   NAME                         READY   UP-TO-DATE   AVAILABLE   AGE
   ceph-rgw-exporter-operator   1/1     1            1           3m

#Verify the auto-discovered exporters

Within about a minute of the Operator starting, one exporter exists for each CephObjectStore.

1. List the exporters.

   kubectl -n rook-ceph get cephrgwexporters

   Expected output (ceph-cluster-object is the object store name in this example):

   NAME                               PHASE     READY   ENDPOINT                                                      AGE
   rgw-exporter-ceph-cluster-object   Running   1       http://rook-ceph-rgw-ceph-cluster-object.rook-ceph.svc:7480   2m

   PHASE is Pending while the exporter Pod starts, Running when it is ready, and Failed if the referenced CephObjectStoreUser or its Secret cannot be resolved. Details are recorded in status.conditions.

2. Confirm the read-only RGW user that was provisioned for the exporter.

   kubectl -n rook-ceph get cephobjectstoreuser rgw-exporter-user-ceph-cluster-object -o jsonpath='{.spec.capabilities}{"\n"}{.status.phase}{"\n"}'

   Expected output:

   {"bucket":"read","metadata":"read","usage":"read","user":"read"}
   Ready

3. Confirm the child resources.

   kubectl -n rook-ceph get deploy,svc,servicemonitor -l app.kubernetes.io/name=ceph-rgw-exporter

#Verify the metrics

1. Forward the exporter Service to a local port and request the metrics endpoint.

   kubectl -n rook-ceph port-forward svc/rgw-exporter-ceph-cluster-object 9242:9242 &
   curl -s http://localhost:9242/-/ready
   curl -s http://localhost:9242/metrics | grep -v '^#' | head

   Expected output:

   ok
   radosgw_usage_ops_total{bucket="bucket_root",category="list_buckets",owner="anonymous",store="ceph-cluster-object"} 142126
   radosgw_usage_scrape_duration_seconds 0.271208467
   radosgw_usage_user_quota_enabled{store="ceph-cluster-object",user="rgw-exporter-user-ceph-cluster-object"} 0
   radosgw_usage_user_total_bytes{store="ceph-cluster-object",user="rgw-exporter-user-ceph-cluster-object"} 0
   ...

   Bucket metrics (radosgw_usage_bucket_*) only appear once at least one bucket exists in the object store.

2. If the monitoring plugin is installed, the ServiceMonitor is discovered automatically: it carries the prometheus: kube-prometheus label that the platform uses to select scrape targets, and scrapes the http-metrics port every 300s by default. After one scrape interval, query radosgw_usage_scrape_duration_seconds in the platform's monitoring dashboards or through PromQL to confirm the data is being ingested. See Management of Metrics for how to work with custom metrics.