GERONIMO-6727 - Issues posting to Zipkin V2 endpoint

[Jonathan Gallimore <jo...@gmail.com>]

Hi Folks

I opened https://issues.apache.org/jira/browse/GERONIMO-6727 and have
created a PR: https://github.com/apache/geronimo-opentracing/pull/1 that I
hope can be considered.

The issue here boils down to 3 things:

1. Using the V1 endpoint works, but data isn't visible in the UI

2. Posting to the V2 endpoint results in a 404 if the Accept:
application/json header is set (the endpoint doesn't return JSON)

3. Posts that include binaryAnnotations in the payload are dropped by the
V2 endpoint as this element is not valid in the V2 payload.

The PR attempts to rectify these by introducing Span objects for V1 and V2,
and dropping the accept header. I have included a test which uses a
modified version of ZipkinRule to start a Zipkin server for the test and
fail in the same way the actual server does is "binaryAnnotations" is
present for a V2 request.

Feedback on the PR is appreciated. Happy to make changes.

Cheers

Jon

Re: GERONIMO-6727 - Issues posting to Zipkin V2 endpoint

[Romain Manni-Bucau <rm...@gmail.com>]

Hi Jon

Commented the PR, hope it makes sense.
Feel free to shout there if anything is unclear.

And a big thanks to cleaning that up!

Ps/tip if it helps: we can also "just" test our models using zipkin
serializers directly, no container needed ;).

Le ven. 3 mai 2019 à 16:23, Jonathan Gallimore <jo...@gmail.com>
a écrit :

> Hi Folks
>
> I opened https://issues.apache.org/jira/browse/GERONIMO-6727 and have
> created a PR: https://github.com/apache/geronimo-opentracing/pull/1 that
> I hope can be considered.
>
> The issue here boils down to 3 things:
>
> 1. Using the V1 endpoint works, but data isn't visible in the UI
>
> 2. Posting to the V2 endpoint results in a 404 if the Accept:
> application/json header is set (the endpoint doesn't return JSON)
>
> 3. Posts that include binaryAnnotations in the payload are dropped by the
> V2 endpoint as this element is not valid in the V2 payload.
>
> The PR attempts to rectify these by introducing Span objects for V1 and
> V2, and dropping the accept header. I have included a test which uses a
> modified version of ZipkinRule to start a Zipkin server for the test and
> fail in the same way the actual server does is "binaryAnnotations" is
> present for a V2 request.
>
> Feedback on the PR is appreciated. Happy to make changes.
>
> Cheers
>
> Jon
>