There is one data point in series where rate/irate/increase gives an unexpected spike. underlying series has 2 points with a small increase.
time series irate=irate(series[1m])
12:15:00 300,000
12:14:58 400,000
12:14:50 399,974
The series is coming from a recording rule. what could be the reason for this abnormality? is there a bad data point in series not visible in range query but irate picks it up?