New Alerting continuously to pending state / NaN

  • What Grafana version and what operating system are you using?
    Grafana 8.3.3 docker

  • What are you trying to achieve?
    My legacy alerts are converted to the new alerting system but they are continuously switching between Pending and OK state as if they are reacting on last / first part of the query.
    E.g. we log statistics every 2 minutes into InfluxDB.
    My alert is based on Mean or Last from past 15 minutes, group by 5 min. interval.
    This results in multiple series for which I then use the Reduce function to have alerts sent for each series.
    However, even if time range is big enough to contain at least 3-6 measurements, some of the series are showing as NaN in the reduce step part of the time. See screenshot. Then few minutes later everything looks fine. This switches every few minutes, as if it has an issue when some series already have a newer measurement but others not.

  • How are you trying to achieve it?

  • What happened?
    Unclear why some series are returning NaN after reduce step, as there are multiple measurements available

  • What did you expect to happen?
    Consistent behavior of the reduce step

  • Can you copy/paste the configuration(s) that you are having problems with?
    This is the InfluxDB query:
    SELECT mean(“system_mgmtcpuusagepcnt”) FROM “http” WHERE $timeFilter GROUP BY time(5m), “customer”, “instance” fill(previous)

  • Did you receive any errors in the Grafana UI or in related logs? If so, please tell us exactly what they were.
    No errors.

  • Did you follow any online instructions? If so, what is the URL?