v0.25.0 otlp example fails

[Taxalo]

Bug Report

Version

0.25.0, latest jaeger (1.59.0)

Platform

macOS

Explaining

opentelemetry-otlp example returns an error and is not consistent generating traces.

executing the example opentelemetry-otlp with cargo run --example opentelemetry-otlp it returns the following:

    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.21s
     Running `target/debug/examples/opentelemetry-otlp`
2024-07-24T18:36:12.618578Z  INFO foo: opentelemetry_otlp: handle foo monotonic_counter.foo=1 key_1="bar" key_2=10
2024-07-24T18:36:12.618741Z  INFO foo: opentelemetry_otlp: histogram example histogram.baz=10
{"resourceMetrics":{"resource":{"attributes":[{"key":"deployment.environment","value":{"stringValue":"develop"}},{"key":"service.version","value":{"stringValue":"0.25.0"}},{"key":"service.name","value":{"stringValue":"tracing-opentelemetry"}}]},"scopeMetrics":[{"scope":{"name":"tracing/tracing-opentelemetry","version":"0.25.0"},"metrics":[{"name":"foo_named","sum":{"dataPoints":[{"attributes":[{"key":"key_1","value":{"stringValue":"bar"}}],"startTime":"2024-07-24 18:36:12.618","time":"2024-07-24 18:36:12.623","startTimeUnixNano":1721846172618688000,"timeUnixNano":1721846172623969000,"value":1}],"aggregationTemporality":"Cumulative","isMonotonic":true}}]}],"schemaUrl":"https://opentelemetry.io/schemas/1.26.0"}}
Other("[ExportErr(Status { code: Unimplemented, message: \", detailed error message: unknown service opentelemetry.proto.collector.metrics.v1.MetricsService\" })]")

The last line contains an unimplemented error, regarding the metrics service.
The traces that this example would generate, are generated RANDOMLY (sometimes appear on jaeger, sometimes don't) when executing the example, even tho we are getting an error.

If we execute this same example in v0.24.0 we get the following:

    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.07s
     Running `target/debug/examples/opentelemetry-otlp`
2024-07-24T18:44:48.482155Z  INFO foo: opentelemetry_otlp: handle foo monotonic_counter.foo=1 key_1="bar" key_2=10
2024-07-24T18:44:48.482344Z  INFO foo: opentelemetry_otlp: histogram example histogram.baz=10
{"resourceMetrics":{"resource":{"attributes":[{"key":"service.version","value":{"stringValue":"0.24.0"}},{"key":"service.name","value":{"stringValue":"tracing-opentelemetry"}},{"key":"deployment.environment","value":{"stringValue":"develop"}}]},"scopeMetrics":[{"scope":{"name":"tracing/tracing-opentelemetry","version":"0.24.0"},"metrics":[{"name":"foo_named","sum":{"dataPoints":[{"attributes":{"key_1":{"stringValue":"bar"}},"startTime":"2024-07-24 18:44:48.482","time":"2024-07-24 18:44:48.494","startTimeUnixNano":1721846688482274000,"timeUnixNano":1721846688494701000,"value":1}],"aggregationTemporality":"Cumulative","isMonotonic":true}}]}],"schemaUrl":"https://opentelemetry.io/schemas/1.25.0"}}

No error, works as expected, and it generates the traces every time we execute it.

If we revert the changes made to the example, and change a little bit the method to work with the new version changes, It will not work as well.

Tried downgrading jaeger, and error keeps happening.

The error seems to be coming from the following code, when an OtelGuard is dropped:

 impl Drop for OtelGuard {
    fn drop(&mut self) {
        if let Err(err) = self.meter_provider.shutdown() { // <- ERROR HAPPENS HERE
            eprintln!("{err:?}");
        }
        opentelemetry::global::shutdown_tracer_provider();
    }
}

[djc]

I probably won't have time to look into this in detail. Would suggest looking through the opentelemetry-otlp changelog in some detail to see if something stands out.

[Taxalo]

Their otlp example returns the same error (https://github.com/open-telemetry/opentelemetry-rust/tree/main/opentelemetry-otlp/examples/basic-otlp).

[djc]

Okay, I recommend you open an issue against the upstream repo, then.