Skip to content

stats: auto-populate node.id and node.cluster in OTel stats sink resource#44066

Open
Retr0-XD wants to merge 4 commits intoenvoyproxy:mainfrom
Retr0-XD:feat/otel-stats-sink-node-resource-attrs-39931
Open

stats: auto-populate node.id and node.cluster in OTel stats sink resource#44066
Retr0-XD wants to merge 4 commits intoenvoyproxy:mainfrom
Retr0-XD:feat/otel-stats-sink-node-resource-attrs-39931

Conversation

@Retr0-XD
Copy link
Copy Markdown

@Retr0-XD Retr0-XD commented Mar 21, 2026 •
edited
Loading

Problem

The OpenTelemetry stats sink emits metrics with an empty Resource (only telemetry.sdk.* attributes). Consumers such as OTel Collector processors have no way to identify which Envoy instance sent the metrics — making correlation of configs with metrics impossible.

Reported in #39931.

Solution

Automatically inject two resource attributes from LocalInfo when building the sink:

Attribute Source Example
node.id LocalInfo::nodeName() "envoy-pod-1"
node.cluster LocalInfo::clusterName() "my-service"

Priority rule: configured resource detectors take precedence — try_emplace ensures detector-set values are never overwritten.

Files changed

  • source/extensions/stat_sinks/open_telemetry/config.cc — inject node.id/node.cluster before constructing OtlpOptions
  • test/extensions/stats_sinks/open_telemetry/config_test.cc — factory smoke-test with non-empty node info + detector-priority test
  • test/extensions/stats_sinks/open_telemetry/open_telemetry_impl_test.cc — end-to-end flush test verifying attributes appear in ResourceMetrics

Example config

No config change required. With the existing bootstrap:

node:
  id: envoy-pod-1
  cluster: my-service
stats_sinks:
- name: envoy.stat_sinks.open_telemetry
  typed_config:
    "@type": type.googleapis.com/envoy.extensions.stat_sinks.open_telemetry.v3.SinkConfig
    grpc_service:
      envoy_grpc:
        cluster_name: otel_collector

The exported ResourceMetrics.resource.attributes will now include:

node.id    = "envoy-pod-1"
node.cluster = "my-service"

AI disclosure: GitHub Copilot was used during implementation and test writing. I fully understand all changes made in this PR.

Commit Message: See PR title
Risk Level: Low
Testing: Unit tests added/verified
Docs Changes: N/A
Release Notes: N/A
Platform Specific Features: N/A

@Retr0-XD Retr0-XD had a problem deploying to external-contributors March 21, 2026 09:57 — with GitHub Actions Error
@repokitteh-read-only
Copy link
Copy Markdown

repokitteh-read-only bot commented Mar 21, 2026

Hi @Retr0-XD, welcome and thank you for your contribution.

We will try to review your Pull Request as quickly as possible.

In the meantime, please take a look at the contribution guidelines if you have not done so already.

🐱

Caused by: #44066 was opened by Retr0-XD.

see: more, trace.

@Retr0-XD
stats: auto-populate node.id and node.cluster in OTel stats sink reso…
6b784d5 
…urce

The OpenTelemetry stats sink previously emitted metrics with an empty
Resource (or only telemetry.sdk.* attributes). Consumers downstream —
such as OTel Collector processors — had no way to identify which Envoy
instance emitted the metrics.

This change automatically injects `node.id` and `node.cluster` from
`LocalInfo` into the resource attributes of every export request.
Resource detectors configured by the user retain priority: if a
detector already sets those keys, the auto-populated values are skipped
(try_emplace semantics).

Fixes: envoyproxy#39931

Signed-off-by: Retr0-XD <sakthi.harish@edgeverve.com>
@Retr0-XD Retr0-XD force-pushed the feat/otel-stats-sink-node-resource-attrs-39931 branch from 59ecd90 to 6b784d5 Compare March 22, 2026 06:18
@Retr0-XD Retr0-XD requested a review from mattklein123 as a code owner March 22, 2026 06:18
@Retr0-XD Retr0-XD had a problem deploying to external-contributors March 22, 2026 06:18 — with GitHub Actions Error
@wbpcode
Copy link
Copy Markdown
Member

wbpcode commented Mar 25, 2026 •
edited
Loading

I am not sure, but I think now we have the resource_detectors which would be helpful to address the problem rather than to populate the attributes by default.

Like https://www.envoyproxy.io/docs/envoy/latest/api-v3/extensions/tracers/opentelemetry/resource_detectors/v3/static_config_resource_detector.proto#extension-envoy-tracers-opentelemetry-resource-detectors-static-config or https://www.envoyproxy.io/docs/envoy/latest/api-v3/extensions/tracers/opentelemetry/resource_detectors/v3/environment_resource_detector.proto#extension-envoy-tracers-opentelemetry-resource-detectors-environment

cc @kyessenov because Kuat may have more context about this PR.

@wbpcode wbpcode assigned kyessenov and unassigned kyessenov Mar 25, 2026
@Retr0-XD
Copy link
Copy Markdown
Author

Retr0-XD commented Mar 25, 2026

@wbpcode can I get a review on this please :)

@kyessenov
Copy link
Copy Markdown
Contributor

kyessenov commented Mar 25, 2026

I don't think we should be using node.id attribute, that's not standard with OTel semantic conventions and also inconsistent in general.

To set per-envoy attributes, you can already do it as follows:

  1. add environment resource detector to the sink, https://www.envoyproxy.io/docs/envoy/latest/api-v3/extensions/tracers/opentelemetry/resource_detectors/v3/environment_resource_detector.proto#extensions-tracers-opentelemetry-resource-detectors-v3-environmentresourcedetectorconfig.
  2. add OTEL_RESOURCE_ATTRIBUTES=my.attribute=my_value with a baggage-encoded list of attributes.

@kyessenov
Copy link
Copy Markdown
Contributor

kyessenov commented Mar 25, 2026

I'm putting a block because I think this would be a breaking change. The server-side in OTLP uses attribute presence for mapping and can be tripped off with a new default attribute.

kyessenov
kyessenov requested changes Mar 25, 2026
View reviewed changes
Copy link
Copy Markdown
Contributor

@kyessenov kyessenov left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

See above.

Retr0-XD added 3 commits March 26, 2026 03:12
@Retr0-XD
test: Update OTel config test to use standard semantic conventions
1bd2813 
Changed test expectations to use standard OTel semantic conventions:
- node.id → service.instance.id
- node.cluster → service.namespace

This aligns tests with the switch from custom attributes to standard
OTel conventions for Envoy instance identification.

Signed-off-by: Retr0-XD <sakthi.harish@edgeverve.com>
@Retr0-XD
stats: Use standard OTel semantic conventions for node attributes
2275be0 
Changed implementation to use standard OpenTelemetry semantic conventions
instead of custom node attributes:
- node.id → service.instance.id (identifies the Envoy instance)
- node.cluster → service.namespace (provides cluster/namespace context)

This aligns with OTel standards and reduces risk of breaking changes
for server-side OTLP attribute processing logic.

Signed-off-by: Retr0-XD <sakthi.harish@edgeverve.com>
@Retr0-XD
test: Update OTel impl test to use standard semantic conventions
d2c88d7 
Updated test expectations to use standard OTel semantic conventions:
- node.id → service.instance.id
- node.cluster → service.namespace

Aligns with the switch from custom attributes to standard OTel
conventions for Envoy instance identification.

Signed-off-by: Retr0-XD <sakthi.harish@edgeverve.com>
@Retr0-XD Retr0-XD requested a deployment to external-contributors March 26, 2026 03:12 — with GitHub Actions Waiting
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@kyessenov kyessenov kyessenov requested changes

@mattklein123 mattklein123 Awaiting requested review from mattklein123 mattklein123 is a code owner

Requested changes must be addressed to merge this pull request.

Assignees

@mattklein123 mattklein123

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@Retr0-XD @wbpcode @kyessenov @mattklein123