[processor/tailsampling] Ability to understand which tailsampling policies attributed to a sampling decision of a trace

[mugli]

Component(s)

processor/tailsampling

Is your feature request related to a problem? Please describe.

We frequently need to understand which tailsampling policies contributed to the decision of sampling a trace. The reasons:

• Understanding contributing factors
• Understanding if we need to optimize certain policies

Additionally, if we could know which traces are sampled from random/probabilistic sampling, we could use those particular subset of traces to do further analytics (calculate distribution of error, latency). It's not possible now because without identifying traces from random/probabilistic sampling the distribution is skewed (towards error, latency etc)

Describe the solution you'd like

Can we make it (optionally) configurable to attach an attribute in the root-span of a trace with the names of root-level policies that evaluated true during tailsampling, also another attribute of list of policies that evaluated to false?

Describe alternatives you've considered

No response

Additional context

No response

[github-actions]

Pinging code owners:

• processor/tailsampling: @jpkrohling

See Adding Labels via Comments if you do not have permissions to add labels yourself.

[mugli]

Sorry, I just realized it might be duplicate of this: #35180

Closing this one. I'll follow the other one as that came first.