[receiver/dockerstats] Dimensions not created without explicit mapping via config.

[austinlparker]

What happened?

Description

When enabling the docker stats receiver, metrics do not receive per-container dimensions by default. I had to add the following to get dimensions:

    container_labels_to_metric_labels:
      com.docker.compose.service: service.name

Steps to Reproduce

Enable dockerstats receiver with no additional config using the opentelemetry demo app.
(Bind the Docker socket into the collector container and set user: "0:0" as well)

Expected Result

Labels would be added for container names.

Actual Result

Labels are not added for container name.

Collector version

0.63.1

Environment information

Environment

OS: macOS Ventura
Running collector-contrib docker container.

OpenTelemetry Collector configuration

receivers:
  otlp:
    protocols:
      grpc:
      http:
        cors:
          allowed_origins:
            - "http://*"
            - "https://*"
  otlp/spanmetrics:
    protocols:
      grpc:
        endpoint: "localhost:65535"
  docker_stats:

exporters:
  otlp:
    endpoint: "jaeger:4317"
    tls:
      insecure: true
  logging:
  prometheus:
    endpoint: "otelcol:9464"


processors:
  batch:
  spanmetrics:
    metrics_exporter: prometheus
  resourcedetection:
    detectors: [docker]

service:
  pipelines:
    traces:
      receivers: [otlp]
      processors: [spanmetrics, batch]
      exporters: [logging, otlp]
    metrics:
      receivers: [otlp, docker_stats]
      processors: [resourcedetection, batch]
      exporters: [prometheus, logging]
    metrics/spanmetrics:
      receivers: [ otlp/spanmetrics ]
      exporters: [ prometheus, logging ]

Log output

Output at /metrics without the additional config:


# HELP container_cpu_percent Percent of CPU used by the container.
# TYPE container_cpu_percent gauge
container_cpu_percent 0.09058823529411765

Output at /metrics with the additional config:

# HELP container_cpu_percent Percent of CPU used by the container.
# TYPE container_cpu_percent gauge
container_cpu_percent{job="adservice"} 0.3043192488262911
container_cpu_percent{job="cartservice"} 2.744461778471139
container_cpu_percent{job="checkoutservice"} 0.03446009389671362
container_cpu_percent{job="currencyservice"} 0.08143525741029642
container_cpu_percent{job="emailservice"} 0.032582159624413146
container_cpu_percent{job="featureflagservice"} 2.6804368174726987
container_cpu_percent{job="ffs_postgres"} 0.448262910798122
container_cpu_percent{job="frontend"} 2.475163806552262
container_cpu_percent{job="frontendproxy"} 0.9003755868544602
container_cpu_percent{job="grafana"} 0.13020280811232451
container_cpu_percent{job="jaeger"} 0.8588087774294672
container_cpu_percent{job="loadgenerator"} 1.353978159126365
container_cpu_percent{job="otelcol"} 4.52371875
container_cpu_percent{job="paymentservice"} 0.06112676056338028
container_cpu_percent{job="productcatalogservice"} 0.21890109890109888
container_cpu_percent{job="prometheus"} 0
container_cpu_percent{job="quoteservice"} 0.02124804992199688
container_cpu_percent{job="recommendationservice"} 2.9113964686998397
container_cpu_percent{job="redis-cart"} 0.43873817034700313
container_cpu_percent{job="shippingservice"} 0

### Additional context

_No response_

[jamesmoessis]

Hey @austinlparker - container.name should be under resource attributes. I'd be interested to see the logging exporter output with loglevel: debug to see if there are resource attributes there.

[austinlparker]

@jamesmoessis I've uploaded logs to this gist: https://gist.github.com/austinlparker/934e375be652b7bf4e07ede4edbfa041

Looking at the prometheus endpoint on the collector, I only see those resource attributes in the target_info metric.

[jamesmoessis]

@austinlparker thanks for uploading the logs - To me it looks like it's working as expected:

Resource attributes:
     -> container.hostname: Str(548565f6b30d)
     -> container.id: Str(548565f6b30ddeba8ba2b62c1f830b46f4f27381c08767a54270fbef81f2464b)
     -> container.image.name: Str(ghcr.io/open-telemetry/demo:v1.0.0-currencyservice)
     -> container.name: Str(currency-service)
     -> container.runtime: Str(docker)
     -> service.name: Str(currencyservice)

The metric you mentioned container.cpu.percent is grouped under that resource.

Perhaps there's something else in your pipeline which is affecting resource attributes. Maybe the prometheus exporter but I'm not familiar with that exporter so can't be sure!

[austinlparker]

Hm, maybe there's something in the Prometheus exporter about applying resource attributes to metrics. I'll have to take a look.

[jamesmoessis]

Ooo yep, you commented back before I edited my comment saying it could be the prom exporter 😆

[jamesmoessis]

For any triager: the tl;dr is that this can be closed. It's not an issue with the dockerstats receiver.

[bravecobra]

Strange as I can't seem to get it to work without adding that extra config. I understand that it should work without that line?

[github-actions]

This issue has been inactive for 60 days. It will be closed in 60 days if there is no activity. To ping code owners by adding a component label, see Adding Labels via Comments, or if you are unsure of which component this issue relates to, please ping @open-telemetry/collector-contrib-triagers. If this issue is still relevant, please ping the code owners or leave a comment explaining why it is still relevant. Otherwise, please close it.

[github-actions]

Pinging code owners for receiver/dockerstats: @rmfitzpatrick @jamesmoessis. See Adding Labels via Comments if you do not have permissions to add labels yourself.