feat(dashboards): Add Node.js runtime metrics prebuilt dashboard UI by chargome · Pull Request #113517 · getsentry/sentry

chargome · 2026-04-21T08:40:08Z

Add the Node.js Runtime Metrics prebuilt dashboard and its onboarding panel. Seven widgets:

Event Loop Utilization, CPU Utilization, Process Uptime (KPIs)
Memory Usage (RSS / heap total / heap used — area chart)
CPU Utilization Over Time (line)
HTTP Request Duration (p50 / p95 — line, spans dataset)

Surfaces the node.runtime.* trace metrics emitted by nodeRuntimeMetricsIntegration (getsentry/sentry-javascript #19923) so users can monitor CPU, memory, and event loop health alongside request latency.

Metric names, types, and aggregations are aligned with the SDK:

max() on the pre-computed p50/p99 event loop delay percentiles (averaging pre-computed percentiles is statistically incorrect).
sum() on process.uptime — the SDK emits elapsed-since-last-tick deltas as a counter, so summing across the window gives total process-seconds.

Trade-off: users with transactions but no runtime metrics will see empty widgets rather than the onboarding panel.

Onboarding:

Dashboard:

depends on #113516
ref getsentry/sentry-javascript#19923
closes https://linear.app/getsentry/issue/DAIN-1532/create-prebuilt-nodejs-health-dashboard

linear-code · 2026-04-21T08:40:13Z

DAIN-1532 Create Prebuilt Node.js Health Dashboard

gggritso

This LGTM overall, but the aggregation of percentiles is … very scary, are you sure you want to do that?

gggritso · 2026-04-22T16:19:54Z

+          name: '',
+          // max() is used because p50/p99 are precomputed percentiles from the SDK's
+          // event loop delay histogram. Averaging precomputed percentiles is statistically
+          // incorrect; max() surfaces the worst-observed value per bucket.
+          fields: [
+            metric('max', 'node.runtime.event_loop.delay.p50', 'gauge', 'second'),
+            metric('max', 'node.runtime.event_loop.delay.p99', 'gauge', 'second'),
+          ],
+          aggregates: [
+            metric('max', 'node.runtime.event_loop.delay.p50', 'gauge', 'second'),
+            metric('max', 'node.runtime.event_loop.delay.p99', 'gauge', 'second'),
+          ],


Oh man! This feels like a statistics crime… am I guessing right that the SDK doesn't emit the actual measurements? What interval is the p50 and p99 collected over? The docs just say "periodically". IMO this is highly spooky. Re-aggregating aggregated metrics in any context is pretty misleading. Imagine someone has ~100 VMs running Node.js, and they're collecting metrics on this. If there were 1,000,000 measurements of 10ms and two measurements of 1,000ms the true p99 is 10ms, but with aggregation-on-aggregation this could spit out something totally different! Is there any other way to do this?

Yeah fair point, this is basically just showing the worst observed percentile since we max() on it instead of avg(). The SDK can't emit raw measurements here but only pre-aggregated values from the node api. I don't really see a solution here as we cannot show this per instance (e.g. in the serverless case). We could either explicitly call this out or just drop the widget wdyt?

🤔 I feel like this data is valuable, and I don't want to hide it completely, it's specifically the aggregation I'm worried about. What do you think about, instead of using a chart, showing a table of measurements? Sentry metrics are individual events, so maybe we can take advantage of that here, and show a table of worst p99 event loop delay measurements?

replaced the widget with two tables for showing the top 10 worst cases for both p50 and p99, no more aggregation on this data

😭 sorry I misled you here, we don't support Table widgets for Metrics datasets! We will soon though, I think, so maybe worth waiting?

oh right, I just ported this over from the custom dashboard where it did work. We can ship without it and create a follow up maybe?

Sounds good to me 👍

gggritso

LGTM, just two nits about using existing types 👍🏻

gggritso · 2026-04-29T13:53:53Z

@@ -0,0 +1,13 @@
+type TraceMetricAggregation = 'avg' | 'sum' | 'max';
+type TraceMetricType = 'gauge' | 'counter';


You can use the existing TraceMetricTypeValue type

gggritso · 2026-04-29T13:54:47Z

@@ -0,0 +1,13 @@
+type TraceMetricAggregation = 'avg' | 'sum' | 'max';
+type TraceMetricType = 'gauge' | 'counter';
+type TraceMetricUnit = 'none' | 'byte' | 'second';


You can use DataUnit for this one 👍🏻

cursor

Cursor Bugbot has reviewed your changes and found 2 potential issues.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 1d159e0. Configure here.}

add dashboard

e4812b2

chargome self-assigned this Apr 21, 2026

chargome requested a review from a team as a code owner April 21, 2026 08:40

github-actions Bot added the Scope: Frontend Automatically applied to PRs that change frontend components label Apr 21, 2026

fix test

fcfb0a9

vercel Bot deployed to Preview April 21, 2026 09:32 View deployment

chargome requested review from bcoe and gggritso April 21, 2026 11:22

gggritso reviewed Apr 22, 2026

View reviewed changes

chargome added 2 commits April 24, 2026 12:03

rm test file

e6e6985

extract trace metric field

bb790f5

vercel Bot deployed to Preview April 24, 2026 10:09 View deployment

event loop delay table

22209fe

vercel Bot deployed to Preview April 27, 2026 08:02 View deployment

cursor Bot reviewed Apr 27, 2026

View reviewed changes

Comment thread static/app/views/dashboards/utils/prebuiltConfigs/nodeRuntimeMetrics/nodeRuntimeMetrics.ts Outdated

chargome and others added 2 commits April 29, 2026 10:31

rm event loop delay tables

5ed23c0

Merge branch 'master' into cg/nodejs-health-dashboard-frontend

46cedba

vercel Bot deployed to Preview April 29, 2026 08:34 View deployment

chargome requested a review from gggritso April 29, 2026 10:02

gggritso reviewed Apr 29, 2026

View reviewed changes

use types

1d159e0

vercel Bot deployed to Preview April 29, 2026 15:30 View deployment

sentry Bot reviewed Apr 29, 2026

View reviewed changes

Comment thread static/app/views/dashboards/utils/prebuiltConfigs/nodeRuntimeMetrics/nodeRuntimeMetrics.ts

cursor Bot reviewed Apr 29, 2026

View reviewed changes

Comment thread static/app/views/dashboards/utils/prebuiltConfigs/utils/traceMetricField.ts Outdated

Comment thread static/app/views/dashboards/utils/prebuiltConfigs/utils/traceMetricField.ts

replace placeholder

184996d

vercel Bot deployed to Preview April 30, 2026 08:59 View deployment

Merge branch 'master' into cg/nodejs-health-dashboard-frontend

86eebba

vercel Bot deployed to Preview April 30, 2026 14:51 View deployment

		@@ -0,0 +1,13 @@
		type TraceMetricAggregation = 'avg' \| 'sum' \| 'max';
		type TraceMetricType = 'gauge' \| 'counter';

Uh oh!

Conversation

chargome commented Apr 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

linear-code Bot commented Apr 21, 2026

Uh oh!

gggritso left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

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!

Uh oh!

gggritso 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!

Uh oh!

cursor Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

chargome commented Apr 21, 2026 •

edited

Loading