State timeline shows brief gaps larger than they should be

[llamafilm]

• What Grafana version and what operating system are you using?
  Grafana Cloud vmain-9553696

I have a state timeline based on a Prometheus query with 1m scrape interval. I configure the panel with “connect null values” <5m in order to mask brief outages, for example if the target was rebooted or Prometheus had a temporary failure. This works well when zoomed in, but it shows unwanted gaps when zoomed out to a larger time range. Does Grafana have a better way to handle this?

One solution could be to increase the query max data points. In this example, 10000 is high enough. But with a 30 time time range, that’s still not enough, and at some point this would run into resource constraints. Is there some intelligent way Grafana could tell Prometheus, “We don’t care about gaps <5m so if you were going to return a null sample, try returning a nearby sample instead” ?

In this example, there is 1 minute gap in data at time 12:00. This first screenshot shows no gap — this is what I want.

This next screenshot shows the same query with a 7 day time range. It shows several breaks in the data, all of which are actually less than 5m.

This is how the data looks in table view:

[llamafilm]

One possible solution: Create separate panels for long term queries, with lower resolution using last_over_time(mymetric[5m]). But this is too much work, not practical to do for every panel at scale. Maybe there could be some way for Grafana to automatically adjust the query like this based on the combination of “connect null values” and the time range?