Add if the trace was sampled to the ctxTags

[domgreen]

Currently we add traceId and spanId to the ctxTags ... which eventually end up on each log line if you are using the ctxTags for logging ... this ends up in a very annoying situation where we have a log line with a span / trace id but when you go hunting for the trace you can never find it.

A fix to this is to add the trace.sampled to the context as well so that we can then identify that the trace will not be in our tracing system.

Previously for a sampled trace would have had :

trace.traceid:123
trace.spanid:987

Which will now be:

trace.traceid:123
trace.spanid:987
trace.sampled:true

Previously for a trace that was not samped would have had :

trace.traceid:456
trace.spanid:654

Which will now be:

trace.traceid:456
trace.spanid:654
trace.sampled:false

[codecov-io]

Codecov Report

Merging #184 into master will decrease coverage by 0.11%.
The diff coverage is 88.88%.

@@            Coverage Diff             @@
##           master     #184      +/-   ##
==========================================
- Coverage    73.2%   73.09%   -0.12%     
==========================================
  Files          36       36              
  Lines        1340     1349       +9     
==========================================
+ Hits          981      986       +5     
- Misses        310      314       +4     
  Partials       49       49

Impacted Files	Coverage Δ
tracing/opentracing/server_interceptors.go	79.06% <100%> (ø)	⬆️
tracing/opentracing/id_extract.go	90.9% <88.23%> (+6.29%)	⬆️
retry/retry.go	76.11% <0%> (-2.23%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 4832df0...71cd4a0. Read the comment docs.

[domgreen]

@devnev ping

[devnev]

How about enforcing that the val is either true or false, and ignoring it if it isn't?

[domgreen]

We could do that however both OpenZipkin and OpenTracing do strconv.FormatBool(sc.Sampled) to set the value

[domgreen]

I think I would prefer to pass thru what is in the sampled field from the OT implementation if it is not true/false it would be valuable to have this visualised.

[domgreen]

@devnev is this change blocking a merge?

[devnev]

Why set to empty value instead of leaving unset?

[domgreen]

😂 this was what you suggested during our conversation ... have updated 👍