Here is what my report looks like.

Use the Dashboard Database so that we can query both the system's 'calendar_date' table and your report's table. This query only pulls data from 2015 and newer (the calendar table goes back to 1980).

SQL:

SELECT c.calendar_date,coalesce(x.Value,0) as 'Value'

FROM calendar_day c

LEFT JOIN (

    SELECT your_date, your_value as Value

    FROM fred_report_data_segment_0.your_data_report) x

ON c.calendar_date = x.your_date

   WHERE c.calendar_date >= '2015-01-01'

ORDER BY c.calendar_date

Here is the joined data on the metric's chart. On 8-21 and 8-27 we chart the inserted zeros.

Note: Using version 3.3's Single Existing Report Data Source (New). You can use the 'Existing Report' or 'Dashboard DB' to query the same data like this:

SELECT measurement_time, count(*)

FROM report_table_name

GROUP BY 1

'measurement_time' is the snapshot date Metric Insights applies when saving the report as a MySQL table. The count represents the numbers of records Metric Insights pulls into the report. If the count is zero than no data has been loaded. If you have false alerts check the timing of data collection.

Remember the measurement_time column only has dates on a go-forward basis. When you create the metric you will only have one data point to start with. This chart tells is the report pulled in 89 records. That is good. All we care about is that there are records at all.

Create the alert rule to fire when there are no records (value =0). You can change the threshold if needed so that you are alerted when there are unusual values (statistical alerting).