retain propagated context for random trace-ids#8263
Conversation
11a2944 to
c4d97f2
Compare
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #8263 +/- ##
=========================================
Coverage 90.29% 90.30%
+ Complexity 7655 7653 -2
=========================================
Files 842 842
Lines 23083 23084 +1
Branches 2317 2317
=========================================
+ Hits 20843 20845 +2
+ Misses 1519 1518 -1
Partials 721 721 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
c4d97f2 to
df726d4
Compare
b6182c1 to
a9d4586
Compare
| isTraceIdRandom = true; | ||
| // Replace parentContext for sampling with one with RANDOM_TRACE_ID bit set | ||
| parentContextForSampler = ROOT_CONTEXT_WITH_RANDOM_TRACE_ID_BIT; | ||
| parentContextForSampler = parentContext.with(RANDOM_TRACE_ID_PRIMORDIAL_SPAN); |
There was a problem hiding this comment.
This means extra allocations for each and every root span, but it seems necessary to be correct.
cc @PeterF778
There was a problem hiding this comment.
I think that root ("invalid") context with propagated values is pretty rare. Perhaps it is worth to optimize for "pure" root context to avoid the extra allocations for the most common case.
jack-berg
left a comment
There was a problem hiding this comment.
Approving and merging before the 1.61.0 release because I view this as a regression bug introduced in the 1.60.0 release and can't release without it.
|
Thank you for your contribution @cleverchuk! 🎉 We would like to hear from you about your experience contributing to OpenTelemetry by taking a few minutes to fill out this survey. |
fixes #8264