OpenTelemetry: fallback to default storage when not on a Vert.x thread #72
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
The VertxContextStorageProvider is chosen by OpenTelemetry via the service loader mechanism. While it is mandatory to use context for storage of spans created on Vert.x threads, there may be parts of the application implemented without Vert.x. In this case, the storage provider fails with NPE. Instead, it is better to fall back to the default storage mechanism, which uses thread local variables. Signed-off-by: Thomas Segismont <[email protected]>
This was referenced Dec 12, 2023
tsegismont
added a commit
to tsegismont/eventing-kafka-broker
that referenced
this pull request
Jan 13, 2025
In Vert.x 4.5.1, there was a fix to use the OTel default context storage when the Vert.x context storage provider is not invoked on a Vert.x thread. See eclipse-vertx/vertx-tracing#72 Since LoomKafkaProducer invokes the tracer on a virtual thread (or a worker thread), the sending task must be wrapped with the OTel current context. Otherwise, tracing data will be lost at this point. OTel provides an ExecutorService that does just that, so this commit refactors the producer to use an executor service instead of a queue plus manual polling. Important: note that with this implementation, a virtual thread is created for each record, which is different from sending them one after the other with a single thread.
matzew
pushed a commit
to matzew/eventing-kafka-broker
that referenced
this pull request
Jan 14, 2025
In Vert.x 4.5.1, there was a fix to use the OTel default context storage when the Vert.x context storage provider is not invoked on a Vert.x thread. See eclipse-vertx/vertx-tracing#72 Since LoomKafkaProducer invokes the tracer on a virtual thread (or a worker thread), the sending task must be wrapped with the OTel current context. Otherwise, tracing data will be lost at this point. OTel provides an ExecutorService that does just that, so this commit refactors the producer to use an executor service instead of a queue plus manual polling. Important: note that with this implementation, a virtual thread is created for each record, which is different from sending them one after the other with a single thread.
knative-prow bot
pushed a commit
to knative-extensions/eventing-kafka-broker
that referenced
this pull request
Jan 14, 2025
* Bump quarkus BOM to get newer vert.x Signed-off-by: Matthias Wessendorf <[email protected]> * Use unwrapped verxInternal's tracer() Signed-off-by: Matthias Wessendorf <[email protected]> * Revert "Use unwrapped verxInternal's tracer()" This reverts commit dfc8b4a. * Fix ReceiverVerticleTracingTest#traceIsPropagated In Vert.x 4.5.1, there was a fix to use the OTel default context storage when the Vert.x context storage provider is not invoked on a Vert.x thread. See eclipse-vertx/vertx-tracing#72 Since LoomKafkaProducer invokes the tracer on a virtual thread (or a worker thread), the sending task must be wrapped with the OTel current context. Otherwise, tracing data will be lost at this point. OTel provides an ExecutorService that does just that, so this commit refactors the producer to use an executor service instead of a queue plus manual polling. Important: note that with this implementation, a virtual thread is created for each record, which is different from sending them one after the other with a single thread. --------- Signed-off-by: Matthias Wessendorf <[email protected]> Co-authored-by: Thomas Segismont <[email protected]>
matzew
added a commit
to matzew/eventing-kafka-broker
that referenced
this pull request
Feb 4, 2025
* Bump quarkus BOM to get newer vert.x Signed-off-by: Matthias Wessendorf <[email protected]> * Use unwrapped verxInternal's tracer() Signed-off-by: Matthias Wessendorf <[email protected]> * Revert "Use unwrapped verxInternal's tracer()" This reverts commit dfc8b4a. * Fix ReceiverVerticleTracingTest#traceIsPropagated In Vert.x 4.5.1, there was a fix to use the OTel default context storage when the Vert.x context storage provider is not invoked on a Vert.x thread. See eclipse-vertx/vertx-tracing#72 Since LoomKafkaProducer invokes the tracer on a virtual thread (or a worker thread), the sending task must be wrapped with the OTel current context. Otherwise, tracing data will be lost at this point. OTel provides an ExecutorService that does just that, so this commit refactors the producer to use an executor service instead of a queue plus manual polling. Important: note that with this implementation, a virtual thread is created for each record, which is different from sending them one after the other with a single thread. --------- Signed-off-by: Matthias Wessendorf <[email protected]> Co-authored-by: Thomas Segismont <[email protected]>
matzew
pushed a commit
to matzew/eventing-kafka-broker
that referenced
this pull request
Apr 2, 2025
In Vert.x 4.5.1, there was a fix to use the OTel default context storage when the Vert.x context storage provider is not invoked on a Vert.x thread. See eclipse-vertx/vertx-tracing#72 Since LoomKafkaProducer invokes the tracer on a virtual thread (or a worker thread), the sending task must be wrapped with the OTel current context. Otherwise, tracing data will be lost at this point. OTel provides an ExecutorService that does just that, so this commit refactors the producer to use an executor service instead of a queue plus manual polling. Important: note that with this implementation, a virtual thread is created for each record, which is different from sending them one after the other with a single thread.
openshift-merge-bot bot
pushed a commit
to openshift-knative/eventing-kafka-broker
that referenced
this pull request
Apr 8, 2025
…1643) * Some generated update for THIRD-PARTY.txt Signed-off-by: Matthias Wessendorf <[email protected]> * Align more of libs from 3.8.6-SP3 release Signed-off-by: Matthias Wessendorf <[email protected]> * LoomKafkaProducer|Consumer let the background thread finish itself (knative-extensions#4013) * Fix ReceiverVerticleTracingTest#traceIsPropagated In Vert.x 4.5.1, there was a fix to use the OTel default context storage when the Vert.x context storage provider is not invoked on a Vert.x thread. See eclipse-vertx/vertx-tracing#72 Since LoomKafkaProducer invokes the tracer on a virtual thread (or a worker thread), the sending task must be wrapped with the OTel current context. Otherwise, tracing data will be lost at this point. OTel provides an ExecutorService that does just that, so this commit refactors the producer to use an executor service instead of a queue plus manual polling. Important: note that with this implementation, a virtual thread is created for each record, which is different from sending them one after the other with a single thread. --------- Signed-off-by: Matthias Wessendorf <[email protected]> Co-authored-by: Marek Schmidt <[email protected]> Co-authored-by: Thomas Segismont <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The VertxContextStorageProvider is chosen by OpenTelemetry via the service loader mechanism.
While it is mandatory to use context for storage of spans created on Vert.x threads, there may be parts of the application implemented without Vert.x.
In this case, the storage provider fails with NPE. Instead, it is better to fall back to the default storage mechanism, which uses thread local variables.