Thresholds

Client-side threshold aggregators can be used to compute the amount of time or the number of occurrences when the metric violated a user-defined threshold.

Lower and upper thresholds can be specified and both of them can be combined to trigger a violation when the series value is outside of the specified range.

max-threshold

Series violates the threshold when its value exceeds the maximum threshold.

min-threshold

Series violates the threshold when its value is below the minimum threshold.

threshold_percent

Percentage of time within the period when the series values did not exceed the threshold (were normal). Computed as sum of violation intervals divided by the period duration.

threshold_count

Number of sequences within the period when the series values exceeded the threshold. Consecutive observations violating the threshold are treated as one sequence.

threshold_duration

Total duration of intervals within the period when the series values exceeded the threshold.

[series]    
  entity = i-fb38c445  
  metric = aws_ec2.statuscheckfailed.average
  statistic = threshold_percent
  period = 1 hour
  max-threshold = 0.1
  alert-expression = value < 100
  alert-style = fill: red; stroke: red

In this example, the underlying series returns 1 in case of an outage, and 0 when the server is available. The max-threshold of 0.1 means that any value greater than 0 is considered a violation.