-
Notifications
You must be signed in to change notification settings - Fork 2
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Support Points aggregator #1
Comments
Also, we should discontinue use of using the gauge metric type for all values, and use the correct value given the record type (e.g. count, histogram, etc.) While it may not matter from the vendor perspective and in writing graphs, the datadog agent collects and reports on these values differently, and can cause incorrect data depending on what the flush intervals are set to e.g. gauge vs. count submission: https://docs.datadoghq.com/developers/metrics/types/?tab=gauge#example The agent will only report the last value for gauge types in a recording window while reporting the sum aggregation for counter types. |
As written to Marwan on Gitter:
|
Oh, you've done it correctly. I'm not sure what I was looking at. |
@zlozano I could point to open-telemetry/oteps#88 on the topic of how to map into Datadog, or to open-telemetry/opentelemetry-specification#382 about configuring OTel itself. |
I tested this out and observations are being aggregated via aggregator.Points and exported as histrograms when i expect them to result in a datadog type of gauge. |
@chrisleavoy in order to do this, we will need to create our own selector which uses the Except instead of fallthrough, we'd select |
@zlozano cheers thanks.. got it working now. |
For now this should be fixed by 7691e46 and the custom aggregator selector that @zlozano mentioned. I think maybe the library should export its own selector so users know they need to use it until open-telemetry/oteps#88 is resolved. Definitely feel free to re-open the issue if anything comes up Thanks! |
We use the exact selector implementation which ultimately surfaces Measures via a Point aggregator.
https://github.com/open-telemetry/opentelemetry-go/blob/master/sdk/metric/selector/simple/simple.go#L110
With this, we can export our raw points as a histogram or timing. It's also important that these types get exported as a histogram, rather than a gauge since the collector aggregations get applied to histograms:
https://github.com/DataDog/datadog-agent/blob/master/pkg/config/config_template.yaml#L128
I propose we amend the current implementation to also include the points aggregator similar to the original otel implementation: https://github.com/open-telemetry/opentelemetry-go/blob/v0.2.3/exporters/metric/internal/statsd/conn.go#L231
The text was updated successfully, but these errors were encountered: