feat(span_processor): add on_ending callback #3010
Conversation
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #3010 +/- ##
=====================================
Coverage 81.0% 81.1%
=====================================
Files 126 126
Lines 24753 24807 +54
=====================================
+ Hits 20070 20126 +56
+ Misses 4683 4681 -2 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
| }; | ||
| for KeyValue { key, value, .. } in span.attributes { | ||
| if let Some(redacted_uri) = obfuscate_http_auth_url(value.as_str().as_ref()) { | ||
| obfuscated_attributes.push((key.clone(), KeyValue::new(key.clone(), redacted_uri))); |
There was a problem hiding this comment.
mm...doesn't looks like the span is mutated with the updated attribute?
There was a problem hiding this comment.
I forgot to add the add_attribute call at the end 🙈
| @@ -105,6 +105,49 @@ async fn router( | |||
| response | |||
| } | |||
|
|
|||
| fn obfuscate_http_auth_url(s: &str) -> Option<String> { | |||
There was a problem hiding this comment.
lets use a simpler example to show the purpose of on_ending?
eg : checking for a specific attribute and if not present, add it.
| /// `on_ending` is called when a `Span` is ending. The end timestamp has already | ||
| /// been computed. | ||
| /// This method is called synchronously within the `Span::end` API, therefore it | ||
| /// should not block or throw an exception. |
There was a problem hiding this comment.
nit: remove the "throw exception" part as it won't fit Rust...
| /// in the order the span processors have been registered, and mutations to the span | ||
| /// will be visible to the next processor. | ||
| /// | ||
| /// The tracer will call `on_ending` for all span processors before calling `on_end` |
There was a problem hiding this comment.
nit: skip details about who is calling. It's good enough to mention that ending is called before end.
Also mention the difference between overriding on_start vs on_ending, as on_start only has those span attributes provide at span_cration time.
| .with_span_processor(SecondProcessor) | ||
| .with_simple_exporter(exporter) | ||
| .build(); | ||
| provider.tracer("test").start("test_span"); |
There was a problem hiding this comment.
test suggestion:
start the span with some attributes.
then add few more attributes to it.
validate that on_start sees only those attributes added at start time.
validate that on_ending sees attributes added at start time, attributes added via span.setattribute, attributes added by any previous processor.
Issues
SpanProcessor API refactor #2940
Changes
Add
on_endingmethod on thrSpanProcessortrait. Spec linkThis method is called during span ending, before
on_end. it is given a mutable span and is very useful to implement obfuscation for instance.It makes possible the use cases that were the motivation for passing mutable data to
on_endwhile respecting the spec.The feature is still marked as Development in the spec so it's kept behind an experimental feature flag.
This PR cherry picked from #2962
Merge requirement checklist
CHANGELOG.mdfiles updated for non-trivial, user-facing changes