Hi,
can anybody help fixing and understanding what is best approach?
Data comes from InfluxDB. Intent is to trigger an alert when a counter (never decreasing, yellow graph) does not increase for a certain time.
The idea is to calculare spread() of the counter every 4m intervale (counter increases every 4m and is sampled every 1m), and pick only the cases where there are more than 1 points (c>1) (to avoid spread(<1 point>)=0, leading to false positives) and min(<spread>)=0.
Below is a practical with false positives:
The the above image:
- is true positive desired alert
- is false positive, likely due to missing (null) data (in
yellow) - is a mistery
Some questions:
a. How comes 3. triggers?
b. How could I avoid 2. false positives, in case counter data is not sampled (tried fill(linear) (Explore data using InfluxQL | InfluxDB OSS 1.8 Documentation) not with great success)
c. What is best way to achieve intent?
Any help is welcome.
Cheers,
Dataselfservice team
