MongoDB Exporter Grafana Dashboards No Longer Showing Metric Data

I have kube-prometheus-stack deployed on my kubernetes cluster. I have percona mongodb server + operator running within this cluster. I have setup mongodb_exporter as a sidecar. I have also installed the dashboards by firstly converting them using

However i can see the dashboard mongodb-instance-summary is not loading any data.

( i can see i have yet to install that one plugin )

In comparison the mongodb-instances-overview dashboard does show data.

  1. Install latest version of percona mongodb operator + server Percona Operator for MongoDB.

  2. Attach mongodb-exporter as a side car using the following params.

          - image: percona/mongodb_exporter:0.40
            name: metrics 
            - name: EXPORTER_USER
                  name: psmdb-db-internal-secrets
                  key: MONGODB_CLUSTER_MONITOR_USER
            - name: EXPORTER_PASS
                  name: psmdb-db-internal-secrets
            - name: POD_IP
                  fieldPath: status.podIP
            - name: MONGODB_URI
              value: "mongodb://$(EXPORTER_USER):$(EXPORTER_PASS)@[$(POD_IP)]:27017"
            args: ["--discovering-mode", "--compatible-mode", "--collect-all", "--mongodb.uri=$(MONGODB_URI)"]
  1. Convert the dashboard and then install onto grafana instance.
python ../dashboards/MongoDB/MongoDB_Instance_Summary.json

I believe there a few other dashboards not displaying data aswell. I have a feeling it must be to do with the scraping/naming of the underlining metrics, given that one dashboard works and the other’s don’t. These dashboards work in our current production version 6 but on this test cluster which uses version 7 and latest percona operator they don’t work.

@Kay_Khan may not cover all cases and options in the dashboards.
I can see you using Grafana 10 or 11 and PBB dashboards are not yet tested with it.

Can you use PMM instead of separated components?

Okay, I understand. Yes we are using Grafana v10.4.1 (d94d597d84)

Switching away from our current monitoring setup using kube-prometheus-stack to PMM is not an option for us.

It’s very likely that the problem lies in the script (which is just a series of regular expressions), so you can first check the Selectors on the dashboards - are they correct? Second, click on Explore on the panels where there is no data and see the formula - this might provide a hint to fix the issue.