Fix tracing module's Jaeger sampling flag

[oleiade]

What?

This Pull Request contains a fix for the sample flags values being incorrect when using the tracing module's Jaeger propagator.

Why?

As reported by @Blinkuu, the sampling option of the tracing module did not set the trace header flags correctly. That led to traces not being sampled, regardless of the options set on the instrumentHTTP called.

This was caused by the constant values defining the sampled/unsampled flags being inverted. This Pull Request sets their values back to their specified values. It also adds tests ensuring that we observe the correct flags from the propagators Propagate methods under various conditions.

Checklist

• I have performed a self-review of my code.
• I have added tests for my changes.
• I have run linter locally (make ci-like-lint) and all checks pass.
• I have run tests locally (make tests) and all tests pass.
• I have commented on my code, particularly in hard-to-understand areas.

Related PR(s)/Issue(s)

#3212

[Blinkuu]

LGTM, thanks!

[codecov-commenter]

Codecov Report

Merging #3251 (7561e06) into master (447b159) will decrease coverage by 0.06%.
Report is 1 commits behind head on master.
The diff coverage is n/a.

❗ Current head 7561e06 differs from pull request most recent head af55151. Consider uploading reports for the commit af55151 to get more accurate results

@@            Coverage Diff             @@
##           master    #3251      +/-   ##
==========================================
- Coverage   73.23%   73.17%   -0.06%     
==========================================
  Files         259      257       -2     
  Lines       19899    19894       -5     
==========================================
- Hits        14573    14558      -15     
- Misses       4402     4407       +5     
- Partials      924      929       +5     

Flag	Coverage Δ
ubuntu	73.17% <ø> (+0.03%)	⬆️
windows	?

Flags with carried forward coverage won't be shown. Click here to find out more.

Files Changed	Coverage Δ
js/modules/k6/experimental/tracing/propagator.go	100.00% <ø> (ø)

... and 7 files with indirect coverage changes

[oleiade]

This is currently red as a result of the #3253 mishap. As soon as v0.45.1 is released, we'll be able to rebase.

[mstoykov]

@oleiade you should rebase on master - v0.45.1 is just to backport the fix and release a version/tag to be used in extensions until 0.46.0 a week from now

[oleiade]

@mstoykov wilco! Thanks for the head-up 🙇🏻

[codebien]

This is a bit strange to me. It is a boolean, isn't possible to have a single value so a single source of truth?

[oleiade]

This is a small simplification. The sampled/not sampled flag is specified as part of a set of flags: a bit within a byte.

As per the specification it could have a range of values. It just so happens that we're only using the sample flag at the moment, and encode it, at the end of the chain as a string, so keeping the constant as a string is a shortcut that avoids adding conversion code.

Happy to change it to numerical values and do the encoding instead, if you prefer; but I would do it in a different PR (this is a bug fix). Just let me know 👍