We are occasionally seeing 0 request count minute-by-minute in cloudflarezone* timeseries as shown in the first screenshot below due to graphql: Internal server error. This PR backfills the missing data in the subsequent interval by taking into account the last minute that it was successful, in the GraphQL query. Request count would appear to have doubled in the next interval, as shown in the second screenshot, but in reality it actually accounts for the missing data in the previous interval. This does not implement a retry mechanism because CloudFlare has imposed a strict rate-limiting.
Description
We are occasionally seeing 0 request count minute-by-minute in cloudflarezone* timeseries as shown in the first screenshot below due to
graphql: Internal server error
. This PR backfills the missing data in the subsequent interval by taking into account the last minute that it was successful, in the GraphQL query. Request count would appear to have doubled in the next interval, as shown in the second screenshot, but in reality it actually accounts for the missing data in the previous interval. This does not implement a retry mechanism because CloudFlare has imposed a strict rate-limiting.