6.7. Ceph monitoring

TargetDown alert reported for ocs-metrics-exporter

Previously, metrics endpoint of the ocs-metrics-exporter used to be unresponsive as persistent volume resync by ocs-metrics-exporter was blocked indefinitely.

With this fix, the blocking operations from persistent volume resync in ocs-metrics-exporter is removed and the metrics endpoint is responsive. Also, the TargetDown alert for ocs-metrics-exporter no longer appears.

(BZ#2168042)

Label references of object bucket claim alerts

Previously, label for the object bucket claim alerts was not displayed correctly as the format for the label-template was wrong. Also, a blank object bucket claim name was displayed and the description text was incomplete.

With this fix, the format is corrected. As a result, the description text is correct and complete with appropriate object bucket claim name.

(BZ#2188032)

Discrepancy in storage metrics

Previously, the capacity of a pool was reported incorrectly as a wrong metrics query was used in the Raw Capacity card in the Block Pool dashboard.

With this fix, the metrics query in the user interface is updated. As a result, the metrics of the total capacity of a block pool is reported correctly.

(BZ#2252035)

Add managedBy label to rook-ceph-exporter metrics and alerts

Previously, the metrics generated by rook-ceph-exporter did not have the managedBy label. So, it was not possible for the OpenShift console user interface to identify from which StorageSystem the metrics are generated.

With this fix, the managedBy label, which has the name of the StorageSystem as a value, is added through the OCS operator to the storage cluster’s Monitoring spec. This spec is read by the Rook operator and it relabels the ceph-exporter’s ServiceMonitor endpoint labels. As a result, all the metrics generated from this exporter will have the new label managedBy.

(BZ#2255491)