feat(rxjava3): add RxJava 3 context propagation instrumentation by wconti27 · Pull Request #11505 · DataDog/dd-trace-java

wconti27 · 2026-05-29T16:25:12Z

🤖 Generated with APM Instrumentation Toolkit

What Does This Do

Adds instrumentation for RxJava 3 (io.reactivex.rxjava3:rxjava) — context propagation across async reactive subscription boundaries for all 5 reactive types.

Motivation

Customer request for RxJava 3 support. RxJava 2 instrumentation already exists; this extends coverage to the v3 API which uses a new package namespace (io.reactivex.rxjava3.core).

Additional Notes

Design: Context propagation only — no new spans created. Mirrors the existing rxjava-2.0 module exactly:

Hooks each reactive type's constructor (capture active context) and subscribe() (wrap observer with tracing wrapper)
Observable, Flowable, Single, Completable, Maybe — all 5 types instrumented
Flowable handles two subscribe() overloads: subscribe(Subscriber) and subscribe(FlowableSubscriber) — the latter is new in RxJava 3 and shares a single Advice class via DYNAMIC typing + TracingSubscriber implementing both interfaces
RxJava3AsyncResultExtension — equivalent of RxJavaAsyncResultExtension in rxjava-2.0, extends spans until reactive chains complete for @WithSpan-annotated methods

Known limitation: Context propagation through Maybe.delay() (computation scheduler) does not produce child spans in the current implementation — the delayed Flowable tests provide equivalent delay coverage. Tracked as a follow-up investigation.

Contributor Checklist

Format the title according to the contribution guidelines
Assign the type: and (comp: or inst:) labels in addition to any other useful labels

Jira ticket: [PROJ-IDENT]

🤖 Generated with APM Instrumentation Toolkit

Remove toolkit-added skills (apm-integrations, datadog-semantics, observability-patterns) and restore .claude to match master branch. Remove the .claude/ .gitignore entry added by mistake.

dd-octo-sts · 2026-05-29T16:54:32Z

🟢 Java Benchmark SLOs — All performance SLOs passed

Suite	Status
Startup	🟢 pass

SLO thresholds are defined here based on automatically generated metrics. A warning is raised when results are within 5% of the threshold.

PR vs. master results

Scenario	Candidate	master	Δ (95% CI of mean)
startup:insecure-bank:iast:Agent	13.94 s	13.90 s	[-0.8%; +1.3%] (no difference)
startup:insecure-bank:tracing:Agent	12.82 s	12.97 s	[-2.5%; +0.1%] (no difference)
startup:petclinic:appsec:Agent	16.63 s	16.60 s	[-1.1%; +1.5%] (no difference)
startup:petclinic:iast:Agent	16.66 s	16.62 s	[-1.1%; +1.5%] (no difference)
startup:petclinic:profiling:Agent	16.58 s	16.57 s	[-1.6%; +1.6%] (no difference)
startup:petclinic:tracing:Agent	15.80 s	15.79 s	[-1.6%; +1.8%] (no difference)

Commit: 66ab94b7 · CI Pipeline · Benchmarking Platform UI

Load and DaCapo benchmarks can be triggered manually in the GitLab pipeline. Results will appear in the Benchmarking Platform UI after completion.

amarziali

The delayedMaybe testing raises an issue highlighting that continuations are getting leaked on the scheduler - it should be excluded from propagation to have a safe instrumentation

amarziali · 2026-06-01T07:04:41Z

+@AutoService(InstrumenterModule.class)
+public final class RxJavaModule extends InstrumenterModule.ContextTracking {
+  public RxJavaModule() {
+    super("rxjava");


add also rxjava-3

amarziali · 2026-06-01T07:06:22Z

+    transformer.applyAdvice(
+        isMethod()
+            .and(named("subscribe"))
+            .and(takesArguments(1))
+            .and(takesArgument(0, named("org.reactivestreams.Subscriber"))),
+        getClass().getName() + "$PropagateParentSpanAdvice");


prise to have spotted both method. However it looks like that this version delegate to the method above that's public final. So that this can be avoided to be instrumented since it only delegates to the latter

amarziali · 2026-06-01T07:07:42Z

+ * to argument slots of either {@code Flowable#subscribe(Subscriber)} or {@code
+ * Flowable#subscribe(FlowableSubscriber)} from the same Advice class.
+ */
+public final class TracingSubscriber<T> implements FlowableSubscriber<T>, Subscriber<T> {


it will only need to implement FlowableSubscriber if we do not instrument anymore the subscribe(Subscriber) version on the FlowableInstrumentation

amarziali · 2026-06-01T07:17:45Z

+    // Delayed operators (Maybe.delay) run on a scheduler thread; spans may outlive the
+    // subscribing scope, causing the pending-trace reference count to go negative when
+    // strictTraceWrites is on.  Mirror RxJava2Test's useStrictTraceWrites() = false.
+    testConfig.strictTraceWrites(false);


This is deliberately cheating on tests. There is indeed an threadpool where we leak continuations and should be excluded from propagating scopes

amarziali · 2026-06-01T07:25:53Z

+  testImplementation project(':dd-java-agent:instrumentation:datadog:tracing:trace-annotation')
+  testImplementation project(':dd-java-agent:instrumentation:opentelemetry:opentelemetry-annotations-1.20')
+
+  testImplementation group: 'io.reactivex.rxjava3', name: 'rxjava', version: '3.1.10'


the minimum version to test is 3.0.0 and not 3.1.10 unless limitations that should be explicitly commented

amarziali · 2026-06-01T07:26:57Z

+  testImplementation project(':dd-java-agent:instrumentation:opentelemetry:opentelemetry-annotations-1.20')
+
+  testImplementation group: 'io.reactivex.rxjava3', name: 'rxjava', version: '3.1.10'
+  testImplementation group: 'io.opentelemetry.instrumentation', name: 'opentelemetry-instrumentation-annotations', version: '1.28.0'


I would also add previous versions of this instrumentation in order to check for overlaps when testing

amarziali · 2026-06-01T07:29:49Z

+    group = "io.reactivex.rxjava3"
+    module = "rxjava"
+    versions = "[3.0.0,)"
+    assertInverse = true


this is not enough. previous versions of rxjava has io.reactivex.rxjava2 group hence a specific fail section is needed here

amarziali · 2026-06-01T07:32:08Z

+  /**
+   * Tracks a known bug: same as {@link #delayedMaybe()} but with two delay/map stages, so the
+   * downstream chain must survive multiple computation-scheduler hops. Re-enable once Maybe
+   * scheduler-hop instrumentation is fixed.
+   */
+  @Disabled(
+      "Known issue: Maybe.delay() loses span context through the computation scheduler — "
+          + "delayed Flowable provides equivalent coverage in the meantime")


This is good to have tracked down the issue in a test. Also, the IA had an idea from where the issue came. This should be addressed by excluding the propagation on the scheduler otherwise we leak continuations and the spans won't close correctly

wconti27 added 5 commits May 29, 2026 10:39

workflow(rxjava3): merge_layers

c9ce2fb

workflow(rxjava3): test_scaffold

d877672

workflow(rxjava3): integration_code

5f40db9

workflow(rxjava3): test:att6:iter1:diagnosis

5b72730

workflow(rxjava3): test:att6:iter2:diagnosis

d0591b2

wconti27 added apm-integration-toolkit PR created by the APM Instrumentation Toolkit ai-generated AI generated code labels May 29, 2026

This comment has been minimized.

Sign in to view

wconti27 added 2 commits May 29, 2026 12:31

workflow(rxjava3): lint_and_fix:att1:iter1:fix_lint_errors

00fa55a

chore: reset .claude to master state

a3779c2

Remove toolkit-added skills (apm-integrations, datadog-semantics, observability-patterns) and restore .claude to match master branch. Remove the .claude/ .gitignore entry added by mistake.

wconti27 force-pushed the apm-ai-toolkit/new_integration/rxjava3/20260529-141514 branch from 97cd6bd to a3779c2 Compare May 29, 2026 16:35

workflow(rxjava3): review_cycle:att2:iter1:batch_fix

66ab94b

amarziali requested changes Jun 1, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(rxjava3): add RxJava 3 context propagation instrumentation#11505

feat(rxjava3): add RxJava 3 context propagation instrumentation#11505
wconti27 wants to merge 8 commits into
masterfrom
apm-ai-toolkit/new_integration/rxjava3/20260529-141514

wconti27 commented May 29, 2026

Uh oh!

This comment has been minimized.

dd-octo-sts Bot commented May 29, 2026 •

edited

Loading

Uh oh!

amarziali left a comment

Uh oh!

amarziali Jun 1, 2026

Uh oh!

amarziali Jun 1, 2026

Uh oh!

amarziali Jun 1, 2026

Uh oh!

amarziali Jun 1, 2026

Uh oh!

amarziali Jun 1, 2026

Uh oh!

amarziali Jun 1, 2026

Uh oh!

amarziali Jun 1, 2026

Uh oh!

amarziali Jun 1, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

wconti27 commented May 29, 2026

What Does This Do

Motivation

Additional Notes

Contributor Checklist

Uh oh!

This comment has been minimized.

dd-octo-sts Bot commented May 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🟢 Java Benchmark SLOs — All performance SLOs passed

Uh oh!

amarziali left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

dd-octo-sts Bot commented May 29, 2026 •

edited

Loading