This is restoring what I believe is previous behavior to send requests to Stackdriver Monitoring in batches. I believe the work to group a set of metrics from a single firehose event introduced this regression, but I'm not completely sure.
This change should significantly (100x) reduce the RPS to Stackdriver Monitoring.
Change PostMetricEvents to collect all provided metrics into a single TimeSeriesRequest
Move out error telemetry to a specialized error object
This is restoring what I believe is previous behavior to send requests to Stackdriver Monitoring in batches. I believe the work to group a set of metrics from a single firehose event introduced this regression, but I'm not completely sure.
This change should significantly (100x) reduce the RPS to Stackdriver Monitoring.
PostMetricEvents
to collect all provided metrics into a singleTimeSeriesRequest