feat(runtime-metrics): sample event loop delay per iteration on Node 26.5+#9273
Merged
Conversation
Contributor
Overall package sizeSelf size: 6.71 MB Dependency sizes| name | version | self size | total size | |------|---------|-----------|------------| | import-in-the-middle | 3.3.1 | 122.62 kB | 438.86 kB | | opentracing | 0.14.7 | 194.81 kB | 194.81 kB | | dc-polyfill | 0.1.11 | 25.74 kB | 25.74 kB |🤖 This report was automatically generated by heaviest-objects-in-the-universe |
🎉 All green!🧪 All tests passed 🎯 Code Coverage (details) 🔗 Commit SHA: 9c89df5 | Docs | Datadog PR Page | Give us feedback! |
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #9273 +/- ##
========================================
Coverage 96.56% 96.57%
========================================
Files 918 918
Lines 121521 121570 +49
Branches 21522 21186 -336
========================================
+ Hits 117347 117402 +55
+ Misses 4174 4168 -6 Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
BenchmarksBenchmark execution time: 2026-07-13 18:36:35 Comparing candidate commit 9c89df5 in PR branch Found 0 performance improvements and 0 performance regressions! Performance is the same for 2318 metrics, 40 unstable metrics.
|
…26.5+
Node 26.5.0 adds monitorEventLoopDelay({ samplePerIteration }), which samples
the event loop delay once per iteration via libuv prepare/check hooks. It is
strictly more accurate than the interval-timer mode and lets us collect
CPU/GC/heap metrics entirely from JS APIs, so on 26.5+ we skip the
@datadog/native-metrics addon for event loop tracking altogether.
- Gate on the runtime version (NODE_MAJOR/NODE_MINOR): older Node silently
ignores the option, so feature detection is not possible.
- Reset the observer with reset() instead of recreating it each flush; this
preserves the samplePerIteration config (recreating dropped it) and avoids a
per-flush allocation.
- Drop the resolution offset from the delay stats in per-iteration mode.
- Add a version-gated test that exercises the real API on Node >= 26.5.0 and
proves the native addon is never started, plus a faked-version test that
covers the gating logic on every runtime.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
pabloerhard
force-pushed
the
pabloerhard/add-per-iteration-metrics
branch
from
July 9, 2026 15:32
475a51b to
a8b238a
Compare
pabloerhard
marked this pull request as ready for review
July 9, 2026 16:03
BridgeAR
reviewed
Jul 13, 2026
BridgeAR
left a comment
BridgeAR
left a comment
Member
There was a problem hiding this comment.
LGTM with two minor suggestions
Comment on lines
206
to
218
| const avg = Math.max(eventLoopDelayObserver.mean - minimum, 0) | ||
| const sum = Math.round(avg * eventLoopDelayObserver.count) | ||
|
|
||
| if (sum !== 0) { | ||
| // Normalize the metrics to the same format as the native metrics. | ||
| const stats = { | ||
| min: Math.max(eventLoopDelayObserver.min - minimum, 0), | ||
| max: Math.max(eventLoopDelayObserver.max - minimum, 0), | ||
| sum, | ||
| total: sum, | ||
| avg, | ||
| count: eventLoopDelayObserver.count, | ||
| p95: Math.max(eventLoopDelayObserver.percentile(95) - minimum, 0), | ||
| } |
Member
There was a problem hiding this comment.
I believe a couple of steps could be skipped in case EVENT_LOOP_SAMPLE_PER_ITERATION_AVAILABLE is available. That way there is less overhead :)
Contributor
Author
There was a problem hiding this comment.
Yes, I did try but benchmarks showed a very minimal improvement (still an improvement though) hence I left the changes as simple as possible.
Member
There was a problem hiding this comment.
Understandable (we do not call these frequent enough to show as actual overhead)
BridgeAR
approved these changes
Jul 14, 2026
juan-fernandez
pushed a commit
that referenced
this pull request
Jul 14, 2026
…26.5+ (#9273) Node 26.5.0 adds monitorEventLoopDelay({ samplePerIteration }), which samples the event loop delay once per iteration via libuv prepare/check hooks. It is strictly more accurate than the interval-timer mode and lets us collect CPU/GC/heap metrics entirely from JS APIs, so on 26.5+ we skip the @datadog/native-metrics addon for event loop tracking altogether. - Gate on the runtime version (NODE_MAJOR/NODE_MINOR): older Node silently ignores the option, so feature detection is not possible. - Reset the observer with reset() instead of recreating it each flush; this preserves the samplePerIteration config (recreating dropped it) and avoids a per-flush allocation. - Drop the resolution offset from the delay stats in per-iteration mode. - Add a version-gated test that exercises the real API on Node >= 26.5.0 and proves the native addon is never started, plus a faked-version test that covers the gating logic on every runtime.
juan-fernandez
pushed a commit
that referenced
this pull request
Jul 14, 2026
…26.5+ (#9273) Node 26.5.0 adds monitorEventLoopDelay({ samplePerIteration }), which samples the event loop delay once per iteration via libuv prepare/check hooks. It is strictly more accurate than the interval-timer mode and lets us collect CPU/GC/heap metrics entirely from JS APIs, so on 26.5+ we skip the @datadog/native-metrics addon for event loop tracking altogether. - Gate on the runtime version (NODE_MAJOR/NODE_MINOR): older Node silently ignores the option, so feature detection is not possible. - Reset the observer with reset() instead of recreating it each flush; this preserves the samplePerIteration config (recreating dropped it) and avoids a per-flush allocation. - Drop the resolution offset from the delay stats in per-iteration mode. - Add a version-gated test that exercises the real API on Node >= 26.5.0 and proves the native addon is never started, plus a faked-version test that covers the gating logic on every runtime.
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
What does this PR do?
Uses Node 26.5.0's
monitorEventLoopDelay({ samplePerIteration })to sample the event loop delay once per iteration (via libuv prepare/check hooks). On runtimes that support it, we now skip the@datadog/native-metricsaddon for event loop tracking entirely and collect CPU/GC/heap metrics from JS APIs.NODE_MAJOR/NODE_MINOR) — older Node silently ignores the option, so feature detection isn't possible.reset()instead of recreating it each flush. This preserves thesamplePerIterationconfiguration (recreating dropped it after the first flush) and avoids a per-flush allocation.Motivation
The per-iteration sampler is strictly more accurate than the interval-timer mode, and lets us drop the native addon dependency for event loop tracking on newer runtimes.
Additional Notes
master; will rebase before merge.samplePerIterationactually shipped in Node v26.5.0 — the version gate depends on it.🤖 Generated with Claude Code