I noticed when using the Redpanda Ops dashboard JSON in Grafana that when I had a cluster that listed over 100 leaderless partitions in rpk cluster health output that the widget in the dashboard thought that there were only 3.
I think there is either a bug in the metrics calculation for the widget, or (as was suggested on Slack) that the server is not reporting enough metrics data when partitions are leaderless to calculate the answer correctly.
I noticed when using the Redpanda Ops dashboard JSON in Grafana that when I had a cluster that listed over 100 leaderless partitions in
rpk cluster healthoutput that the widget in the dashboard thought that there were only 3. I think there is either a bug in the metrics calculation for the widget, or (as was suggested on Slack) that the server is not reporting enough metrics data when partitions are leaderless to calculate the answer correctly.