fix asynchonous unary call traces

[sengjea]

Description

When calling the GRPC client with future(), the client fails as the context object is re-entered during the asynchonous callback. See test_unary_unary_future. This PR fixes that by using lower level function calls to control the re-entry of context during the synchronous and asynchonous bits.

Type of change

Bug fix

How Has This Been Tested?

• tox -e test-instrumentation-grpc test_client_interceptor.py::TestClientProto::test_unary_unary_future

Does This PR Require a Core Repo Change?

No.

Checklist:

See contributing.md for styleguide, changelog guidelines, and more.

Followed the style guidelines of this project
Unit tests have been added

[linux-foundation-easycla]

The committers are authorized under a signed CLA.

• ✅ Seng Jea Lee (691bac3)

[aabmass]

Nice LGTM overall, but would be great if someone else can also take a look.

intercept_unary() and intercept_stream() seem to duplicate a lot of code, would it be reasonable to refactor them to share it?

[sengjea]

Nice LGTM overall, but would be great if someone else can also take a look.

intercept_unary() and intercept_stream() seem to duplicate a lot of code, would it be reasonable to refactor them to share it?

ok i've refactored some of the code but I think it's getting into can-o-worms territory. There's stuff like RpcInfo which seems like an overhang from the opentracing era where span decorator functions could be passed into the instrumentor for additional span processing. Will do a bigger refactor in another PR.

[aabmass]

ok i've refactored some of the code but I think it's getting into can-o-worms territory. There's stuff like RpcInfo which seems like an overhang from the opentracing era where span decorator functions could be passed into the instrumentor for additional span processing. Will do a bigger refactor in another PR.

Nice thank you for the contribution!

[sengjea]

the docs check is failing because it was unable to reach some endpoint.