Ramji

I'm running into this with `cri-o 1.21` and `k8s 1.23`. The node is showing this error after it successfully scheduled pods. So far, it seems unpredictable when the node will...

@TommyCpp I agree users should be able to enable both `http-proto` and `http-json` features on the crate. I'll test it and perhaps use the protocol to decide on certain things...

@TommyCpp are there existing tests covering the function `build_body` in http exporters?

The latest failures do not show up on pre-commit tests locally. ``` running 11 tests test exporter::config::agent::tests::set_socket_address ... ok test exporter::config::collector::http_client::collector_client_tests::test_bring_your_own_client ... ignored test exporter::config::collector::tests::test_collector_exporter ... ignored test exporter::config::collector::tests::test_resolve_endpoint ......

Pulling back from review to fix test failures when running `cargo test --all`.

@TommyCpp please take a look when you get a chance.

> Overall looks good. My concern is mostly around `http-json` overriding with `http-proto` when both are enabled That's where I had most trouble changing the code. It seems that having...

>I think it's reasonable to rely on features This is still a bit tricky when both `http-json` and `http-proto` features are enabled. Are we ok to pick `http-json` in that...

@TommyCpp sorry, just got to this. Please take a look when you get a chance. Thanks.

Any chance this will get prioritized? This effectively blocks consumers of `opentelemetry_http` crate from moving up to hyper 1.0. In my case, we already moved up and am now blocked...