Closed franciscovalentecastro closed 6 months ago
I assume this is a bug with the current collector implementation. In that case, we would need to merge buckets (scale down, I think?) to get below the upper bound.
After debugging further found that the way i was translating between the OTel ExponetialHistogram
to Cloud Monitoring Distribution
was adding a lot unneeded (empty) buckets to account for the Offset
parameter. By adding Offset
correctly encoded inside the scale parameter ( base ** Offset
) this shouldn't happen anymore, since the MaxSize
number of buckets is 160
.
Closing this one since it was a bug in the translation.
While working on https://github.com/GoogleCloudPlatform/opentelemetry-operations-go/pull/777 found that there is an upper bound of number of exponential buckets that can be easily exceeded when exporting an
Exponential Histogram
.