Metrics that do not have a complete history are not shown at all
There are some metrics where there may not be data for the entire duration. For example if you are trying to show an error rate, you will be dividing by the total number of requests. (HTTP 500's / All Responses)
If there is a period where there are no responses, like a test environment, you are diving by zero which means there will be no data for that time period. We should still show the data for data points that we do have, like Grafana or Prometheus itself.
For example, here is the Prometheus chart for the same query:
And the GitLab chart, note there is no lines or even a Y-axis scale:
The query: sum(rate(nginx_responses_total{status_code="5xx", %{environment_filter}}[2m])) / sum(rate(nginx_responses_total{server_zone!="*", server_zone!="_", %{environment_filter}}[2m])) * 100