We have a datadog metric that measures a % of successful web requests and alerts us if it's more than threshold. A problem with this metric is that it becomes noisy on weekend nights when we don't have too many requests and even a single error can push the metric over the threshold.
Right now the query is like this:
"query": "sum(last_30m):sum:q.inquiry{success:true}.as_count() / sum:q.inquiry.as_count() * 100 < 80"
which counts the number of requests that succeed vs total number of requests and reports if it's more than 80%.
Is there a way to do a boolean operation to do something like
above_query && sum(last_30m):sum:q.inquiry{success:false}.as_count() > 3
which will only alert if the number of bad requests is greater than 3