I have data with irregular timestamps but they are approximately every 15 minutes, and need to query Influx to retrieve the data to create a chart with 100 samples over a given duration of time.
If the desired duration is 100 days, I simply group by 1d and return the mean value, and since there are about 96 samples (1 day / 15 minutes) in each group, the chart is pretty accurate and appears very fluid.
If on the other hand the desired duration is 6 hours, I would need to group by 216 seconds (6h * 60m/h * 60s/m / 100), and since this interval is less than my actual data frequency, I might not even have a data point in a given interval. I tried to overcome this by using group fill of previous or linier, but the chart software (Highcharts) renders the 100 samples as if they were gospel and and the chart looks like crap.
Not sure whether there is anything I could do about this. Any ideas? Thanks