Usage of MockZipkinCollector in integration tests

[pellared]

Currently, the integration tests are using MockZipkinCollector.

1. Should we test against OTLP instead of Zipkin and create a MockOtlpCollector?
2. Should we test against all supported exporters in the integration tests?
3. Is it needed for the beta release?

[pellared]

My opinion

ad. 1. Yes. OTLP exporter does not do any transformations so the assertions would be more trustworthy. ad. 2. We can have only some integration tests to test all of the exporters. However, we should not run all the tests against all possible exporteres. ad. 3. No. It would postpone getting initial feedback.

[nrcventura]

@pellared I agree with your opinions. For 1. We'll eventually need to decide if we want the tests to be grpc based, or http based. For 2. I think that we just need end to end tests just so that we have some confidence that we are bootstrapping all of the correct dependencies.

[pellared]

We'll eventually need to decide if we want the tests to be grpc based, or http based.

I strongly prefer to use grpc as this is the default in https://github.com/open-telemetry/opentelemetry-dotnet/tree/main/src/OpenTelemetry.Exporter.OpenTelemetryProtocol

This issue may be required for beta if we agree on: https://github.com/open-telemetry/opentelemetry-dotnet-instrumentation/issues/255#issuecomment-970116850

[pellared]

We ended up in http/protobuf per https://github.com/open-telemetry/opentelemetry-dotnet-instrumentation/issues/325

[RassK]

@pellared this seems done now?

[pellared]

@RassK Yes. Thanks.