Example to show how to achieve more consistent sampling across linked traces

[kalyanaj]

This is an example to show how to get more consistent sampling even across linked traces. Please see tradeoffs listed in README.md for when this approach can be helpful and when it won't help with consistent sampling across linked traces. This uses a composite sampler that includes a parent based sampler and a links based sampler.

Fixes #4345
Design discussion issue #

Changes

This shows the example of a composite sampler that has:

1. A parent based sampler.
2. A links based sampler.

This composite sampler first delegates to the parent based sampler. If the parent based sampler decides to sample, then it decides to sample. However, if the parent based sampler decides to drop, the composite sampler delegates to the links based sampler. The links based sampler decides to sample if the activity has any linked activities and if at least ONE of those linked activities is sampled.

Merge requirement checklist

• CONTRIBUTING guidelines followed (nullable enabled, static analysis, etc.)
• Unit tests added/updated
• Appropriate CHANGELOG.md files updated for non-trivial changes
• Changes in public API reviewed (if applicable)

[reyang]

I've left some non-blocking comments.

[codecov]

Codecov Report

Merging #4346 (8d69bcb) into main (678e9c4) will increase coverage by 0.04%.
The diff coverage is n/a.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4346      +/-   ##
==========================================
+ Coverage   84.73%   84.78%   +0.04%     
==========================================
  Files         300      300              
  Lines       12064    12064              
==========================================
+ Hits        10223    10229       +6     
+ Misses       1841     1835       -6     

see 4 files with indirect coverage changes

[pyohannes]

I added two small nits, but no blocker for merging this.

[kalyanaj]

@jmacd FYI on this PR (would love to get your review/feedback on this).

[github-actions]

This PR was marked stale due to lack of activity and will be closed in 7 days. Commenting or Pushing will instruct the bot to automatically remove the label. This bot runs once per day.

[oertl]

This proposal would also fit with consistent probability sampling, if the composite sampler delegates to a ParentConsistentProbabilityBased sampler and an Always-on consistent probability sampler depending on whether there are linked activities.

[kalyanaj]

This proposal would also fit with consistent probability sampling, if the composite sampler delegates to a ParentConsistentProbabilityBased sampler and an Always-on consistent probability sampler depending on whether there are linked activities.

Thanks @oertl . Agree it would fit with it. However, since the consistent probability sampling spec is still experimental, wanted to scope this example to the current stable parent based sampling mechanism. In the future, once the consistent probability sampling mechanism becomes stable, we can update this to reflect how this would work with consistent probability sampling. Please let me know if you feel differently. Thanks!

[utpilla]

I have left some non-blocking suggestions.

[kalyanaj]

@jmacd FYI in case you want to review this as well.

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: kalyanaj / name: J. Kalyana Sundaram (de04538, 4d74805, 322c483, ae25ae8, c084893, f2dfa95, 949420e, 8cda1d2, f483188, 109079f, 32b8238, ec3d1de, 43bc608, 8ff1853, 4cb6d44, 8d69bcb)
• ✅ login: cijothomas / name: Cijo Thomas (c8076b0, 43b07fe)
• ✅ login: utpilla / name: Utkarsh Umesan Pillai (04039e0, 9d06f90, 46b71df)

[jarias-lfx]

/easycla