Deploy and use the collector

This feature is subject to the "Pre-GA Offerings Terms" in the General Service Terms section of the Service Specific Terms. Pre-GA features are available "as is" and might have limited support. For more information, see the launch stage descriptions.

This document describes how to deploy the OpenTelemetry Collector, configure the collector to use the otlphttp exporter and the Telemetry (OTLP) API, and run a telemetry generator to write metrics to Cloud Monitoring. You can then view these metrics in Cloud Monitoring.

If you are using Google Kubernetes Engine, you can follow Managed OpenTelemetry for GKE, instead of manually deploying and configuring an OpenTelemetry Collector that uses the Telemetry API.

If you are using an SDK to send metrics from an application directly to the Telemetry API, then see Use SDKs to send metrics from applications for additional information and examples.

You can also use an OpenTelemetry Collector and the Telemetry API in conjunction with OpenTelemetry zero-code instrumentation. For more information, see Use OpenTelemetry zero-code instrumentation for Java.

Before you begin

This section describes how to set up your environment for deploying and using the collector.

Select or create a Google Cloud project

Choose a Google Cloud project for this walkthrough. If you don't already have a Google Cloud project, then create one:

Sign in to your Google Cloud account. If you're new to Google Cloud, create an account to evaluate how our products perform in real-world scenarios. New customers also get $300 in free credits to run, test, and deploy workloads.

In the Google Cloud console, on the project selector page, select or create a Google Cloud project.

Roles required to select or create a project

• Select a project: Selecting a project doesn't require a specific IAM role—you can select any project that you've been granted a role on.
• Create a project: To create a project, you need the Project Creator role (roles/resourcemanager.projectCreator), which contains the resourcemanager.projects.create permission. Learn how to grant roles.

Note: If you don't plan to keep the resources that you create in this procedure, create a project instead of selecting an existing project. After you finish these steps, you can delete the project, removing all resources associated with the project.

Go to project selector

Verify that billing is enabled for your Google Cloud project.

In the Google Cloud console, on the project selector page, select or create a Google Cloud project.

Roles required to select or create a project

• Select a project: Selecting a project doesn't require a specific IAM role—you can select any project that you've been granted a role on.
• Create a project: To create a project, you need the Project Creator role (roles/resourcemanager.projectCreator), which contains the resourcemanager.projects.create permission. Learn how to grant roles.

Note: If you don't plan to keep the resources that you create in this procedure, create a project instead of selecting an existing project. After you finish these steps, you can delete the project, removing all resources associated with the project.

Go to project selector

Verify that billing is enabled for your Google Cloud project.

Install command-line tools

This document uses following command-line tools:

• gcloud
• kubectl

The gcloud and kubectl tools are part of the Google Cloud CLI. For information about installing them, see Managing Google Cloud CLI components. To see the gcloud CLI components you have installed, run the following command:

gcloud components list

To configure the gcloud CLI for use, run the following commands:

gcloud auth login
gcloud config set project PROJECT_ID

Enable APIs

Enable the Cloud Monitoring API and the Telemetry API in your Google Cloud project. Pay particular attention to the Telemetry API, telemetry.googleapis.com; this document might be the first time you've encountered this API.

Enable the APIs by running the following commands:

gcloud services enable monitoring.googleapis.com
gcloud services enable telemetry.googleapis.com

Create a cluster

Create a GKE cluster.

1. Creates a Google Kubernetes Engine cluster named otlp-test by running the following command:

   gcloud container clusters create-auto --location CLUSTER_LOCATION otlp-test --project PROJECT_ID
   

2. After the cluster is created, connect to it by running the following command:

   gcloud container clusters get-credentials otlp-test --region CLUSTER_LOCATION --project PROJECT_ID
   

Authorize the Kubernetes service account

The following commands grant the necessary Identity and Access Management (IAM) roles to the Kubernetes service account. These commands assume you are using Workload Identity Federation for GKE:

export PROJECT_NUMBER=$(gcloud projects describe PROJECT_ID --format="value(projectNumber)")

gcloud projects add-iam-policy-binding projects/PROJECT_ID \
  --role=roles/logging.logWriter \
  --member=principal://iam.googleapis.com/projects/$PROJECT_NUMBER/locations/global/workloadIdentityPools/PROJECT_ID.svc.id.goog/subject/ns/opentelemetry/sa/opentelemetry-collector \
  --condition=None

gcloud projects add-iam-policy-binding projects/PROJECT_ID \
  --role=roles/monitoring.metricWriter \
  --member=principal://iam.googleapis.com/projects/$PROJECT_NUMBER/locations/global/workloadIdentityPools/PROJECT_ID.svc.id.goog/subject/ns/opentelemetry/sa/opentelemetry-collector \
  --condition=None

gcloud projects add-iam-policy-binding projects/PROJECT_ID \
  --role=roles/telemetry.tracesWriter \
  --member=principal://iam.googleapis.com/projects/$PROJECT_NUMBER/locations/global/workloadIdentityPools/PROJECT_ID.svc.id.goog/subject/ns/opentelemetry/sa/opentelemetry-collector \
  --condition=None

If your service account has a different format, then you can use the command in the Google Cloud Managed Service for Prometheus documentation to authorize the service account, with the following changes:

• Replace the service-account name gmp-test-sa with your service account.
• Grant the roles shown in the previous set of commands, not just the roles/monitoring.metricWriter role.

Deploy the OpenTelemetry Collector

Create the collector configuration by making a copy of the following YAML file and placing it in a file named collector.yaml. You can also find the following configuration on GitHub in the otlp-k8s-ingest repository.

In your copy, be sure to replace the occurrence of ${GOOGLE_CLOUD_PROJECT} with your project ID, PROJECT_ID.

OTLP for Prometheus metrics only works when using the OpenTelemetry Collector version 0.140.0 or newer.

# Copyright 2024 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

exporters:
  # The googlecloud exporter is used for logs
  googlecloud:
    log:
      default_log_name: opentelemetry-collector
    user_agent: Google-Cloud-OTLP manifests:0.4.0 OpenTelemetry Collector Built By Google/0.128.0 (linux/amd64)
  googlemanagedprometheus:
    user_agent: Google-Cloud-OTLP manifests:0.4.0 OpenTelemetry Collector Built By Google/0.128.0 (linux/amd64)
  # The otlphttp exporter is used to send traces to Google Cloud Trace and
  # metrics to Google Managed Prometheus using OTLP http/proto.
  # The otlp exporter could also be used to send them using OTLP grpc
  otlphttp:
    encoding: proto
    endpoint: https://telemetry.googleapis.com
    # Use the googleclientauth extension to authenticate with Google credentials
    auth:
      authenticator: googleclientauth


extensions:
  # Standard for the collector. Used for probes.
  health_check:
    endpoint: ${env:MY_POD_IP}:13133
  # This is an auth extension that adds Google Application Default Credentials to http and gRPC requests.
  googleclientauth:


processors:
  # This filter is a standard part of handling the collector's self-observability metrics. Not related to OTLP ingestion.
  filter/self-metrics:
    metrics:
      include:
        match_type: strict
        metric_names:
        - otelcol_process_uptime
        - otelcol_process_memory_rss
        - otelcol_grpc_io_client_completed_rpcs
        - otelcol_googlecloudmonitoring_point_count

  # The recommended batch size for the OTLP endpoint is 200 metric data points.
  batch:
    send_batch_max_size: 200
    send_batch_size: 200
    timeout: 5s

  # The k8sattributes processor adds k8s resource attributes to metrics based on the source IP that sent the metrics to the collector.
  # k8s attributes are important for avoiding errors from timeseries "collisions".
  # These attributes help distinguish workloads from each other, and provide useful metadata (e.g. namespace) when querying.
  k8sattributes:
    extract:
      metadata:
      - k8s.namespace.name
      - k8s.deployment.name
      - k8s.statefulset.name
      - k8s.daemonset.name
      - k8s.cronjob.name
      - k8s.job.name
      - k8s.replicaset.name
      - k8s.node.name
      - k8s.pod.name
      - k8s.pod.uid
      - k8s.pod.start_time
    passthrough: false
    pod_association:
    - sources:
      - from: resource_attribute
        name: k8s.pod.ip
    - sources:
      - from: resource_attribute
        name: k8s.pod.uid
    - sources:
      - from: connection

  # Standard processor for gracefully degrading when overloaded to prevent OOM.
  memory_limiter:
    check_interval: 1s
    limit_percentage: 65
    spike_limit_percentage: 20

  # Standard processor for enriching self-observability metrics. Unrelated to OTLP ingestion.
  metricstransform/self-metrics:
    transforms:
    - action: update
      include: otelcol_process_uptime
      operations:
      - action: add_label
        new_label: version
        new_value: Google-Cloud-OTLP manifests:0.4.0 OpenTelemetry Collector Built By Google/0.128.0 (linux/amd64)

  # The resourcedetection processor, similar to the k8sattributes processor, enriches metrics with important metadata.
  # The gcp detector provides the cluster name and cluster location.
  resourcedetection:
    detectors: [gcp]
    timeout: 10s

  # This transform processor avoids ingestion errors if metrics contain attributes with names that are reserved for the prometheus_target resource.
  transform/collision:
    metric_statements:
    - context: datapoint
      statements:
      - set(attributes["exported_location"], attributes["location"])
      - delete_key(attributes, "location")
      - set(attributes["exported_cluster"], attributes["cluster"])
      - delete_key(attributes, "cluster")
      - set(attributes["exported_namespace"], attributes["namespace"])
      - delete_key(attributes, "namespace")
      - set(attributes["exported_job"], attributes["job"])
      - delete_key(attributes, "job")
      - set(attributes["exported_instance"], attributes["instance"])
      - delete_key(attributes, "instance")
      - set(attributes["exported_project_id"], attributes["project_id"])
      - delete_key(attributes, "project_id")

  # The relative ordering of statements between ReplicaSet & Deployment and Job & CronJob are important.
  # The ordering of these controllers is decided based on the k8s controller documentation available at
  # https://kubernetes.io/docs/concepts/workloads/controllers.
  # The relative ordering of the other controllers in this list is inconsequential since they directly
  # create pods.
  transform/aco-gke:
    metric_statements:
    - context: datapoint
      statements:
      - set(attributes["top_level_controller_type"], "ReplicaSet") where resource.attributes["k8s.replicaset.name"] != nil
      - set(attributes["top_level_controller_name"], resource.attributes["k8s.replicaset.name"]) where resource.attributes["k8s.replicaset.name"] != nil
      - set(attributes["top_level_controller_type"], "Deployment") where resource.attributes["k8s.deployment.name"] != nil
      - set(attributes["top_level_controller_name"], resource.attributes["k8s.deployment.name"]) where resource.attributes["k8s.deployment.name"] != nil
      - set(attributes["top_level_controller_type"], "DaemonSet") where resource.attributes["k8s.daemonset.name"] != nil
      - set(attributes["top_level_controller_name"], resource.attributes["k8s.daemonset.name"]) where resource.attributes["k8s.daemonset.name"] != nil
      - set(attributes["top_level_controller_type"], "StatefulSet") where resource.attributes["k8s.statefulset.name"] != nil
      - set(attributes["top_level_controller_name"], resource.attributes["k8s.statefulset.name"]) where resource.attributes["k8s.statefulset.name"] != nil
      - set(attributes["top_level_controller_type"], "Job") where resource.attributes["k8s.job.name"] != nil
      - set(attributes["top_level_controller_name"], resource.attributes["k8s.job.name"]) where resource.attributes["k8s.job.name"] != nil
      - set(attributes["top_level_controller_type"], "CronJob") where resource.attributes["k8s.cronjob.name"] != nil
      - set(attributes["top_level_controller_name"], resource.attributes["k8s.cronjob.name"]) where resource.attributes["k8s.cronjob.name"] != nil
  # For each Prometheus unknown-typed metric, which is a gauge, create a counter that is an exact copy of this metric.
  # The GCP OTLP endpoint will add appropriate the appropriate suffixes for the counter and gauge.
  transform/unknown-counter:
    metric_statements:
    - context: metric
      statements:
      # Copy the unknown metric, but add a suffix so we can distinguish the copy from the original.
      - copy_metric(Concat([metric.name, "unknowncounter"], ":")) where metric.metadata["prometheus.type"] == "unknown" and not HasSuffix(metric.name, ":unknowncounter")
      # Change the copy to a monotonic, cumulative sum.
      - convert_gauge_to_sum("cumulative", true) where HasSuffix(metric.name, ":unknowncounter")
      # Delete the extra suffix once we are done.
      - set(metric.name, Substring(metric.name, 0, Len(metric.name)-Len(":unknowncounter"))) where HasSuffix(metric.name, ":unknowncounter")

  # When sending telemetry to the GCP OTLP endpoint, the gcp.project_id resource attribute is required to be set to your project ID.
  resource/gcp_project_id:
    attributes:
    - key: gcp.project_id
      # MAKE SURE YOU REPLACE THIS WITH YOUR PROJECT ID
      value: ${GOOGLE_CLOUD_PROJECT}
      action: insert
  # The metricstarttime processor is important to include if you are using the prometheus receiver to ensure the start time is set properly.
  # It is a no-op otherwise.
  metricstarttime:
    strategy: subtract_initial_point

receivers:
  # This collector is configured to accept OTLP metrics, logs, and traces, and is designed to receive OTLP from workloads running in the cluster.
  otlp:
    protocols:
      grpc:
        endpoint: ${env:MY_POD_IP}:4317
      http:
        cors:
          allowed_origins:
          - http://*
          - https://*
        endpoint: ${env:MY_POD_IP}:4318

  # Push the collector's own self-observability metrics to the otlp receiver.
  otlp/self-metrics:
    protocols:
      grpc:
        endpoint: ${env:MY_POD_IP}:14317

service:
  extensions:
  - health_check
  - googleclientauth
  pipelines:
    # Recieve OTLP logs, and export logs using the googlecloud exporter.
    logs:
      exporters:
      - googlecloud
      processors:
      - k8sattributes
      - resourcedetection
      - memory_limiter
      - batch
      receivers:
      - otlp
    # Recieve OTLP metrics, and export metrics to GMP using the otlphttp exporter.
    metrics/otlp:
      exporters:
      - otlphttp
      processors:
      - k8sattributes
      - memory_limiter
      - resource/gcp_project_id
      - resourcedetection
      - transform/collision
      - transform/aco-gke
      - transform/unknown-counter
      - metricstarttime
      - batch
      receivers:
      - otlp
    # Scrape self-observability Prometheus metrics, and export metrics to GMP using the otlphttp exporter.
    metrics/self-metrics:
      exporters:
      - otlphttp
      processors:
      - filter/self-metrics
      - metricstransform/self-metrics
      - k8sattributes
      - memory_limiter
      - resource/gcp_project_id
      - resourcedetection
      - batch
      receivers:
      - otlp/self-metrics
    # Recieve OTLP traces, and export traces using the otlphttp exporter.
    traces:
      exporters:
      - otlphttp
      processors:
      - k8sattributes
      - memory_limiter
      - resource/gcp_project_id
      - resourcedetection
      - batch
      receivers:
      - otlp
  telemetry:
    logs:
      encoding: json
    metrics:
      readers:
      - periodic:
          exporter:
            otlp:
              protocol: grpc
              endpoint: ${env:MY_POD_IP}:14317

Configure the deployed OpenTelemetry Collector

Configure the collector deployment by creating Kubernetes resources.

1. Create the opentelemetry namespace and create the collector configuration in the namespace by running the following commands:

   kubectl create namespace opentelemetry
   
   kubectl create configmap collector-config -n opentelemetry --from-file=collector.yaml
   

2. Configure the collector with Kubernetes resources by running the following commands:

   kubectl apply -f https://raw.githubusercontent.com/GoogleCloudPlatform/otlp-k8s-ingest/refs/heads/otlpmetric/k8s/base/2_rbac.yaml
   
   kubectl apply -f https://raw.githubusercontent.com/GoogleCloudPlatform/otlp-k8s-ingest/refs/heads/otlpmetric/k8s/base/3_service.yaml
   
   kubectl apply -f https://raw.githubusercontent.com/GoogleCloudPlatform/otlp-k8s-ingest/refs/heads/otlpmetric/k8s/base/4_deployment.yaml
   
   kubectl apply -f https://raw.githubusercontent.com/GoogleCloudPlatform/otlp-k8s-ingest/refs/heads/otlpmetric/k8s/base/5_hpa.yaml
   

3. Wait for the collector pods to reach "Running" and have 1/1 containers ready. This takes about three minutes on Autopilot, if this is the first workload deployed. To check the pods, use the following command:

   kubectl get po -n opentelemetry -w
   

   To stop watching the pod status, enter Ctrl-C to stop the command.

4. You can also check the collector logs to make sure there aren't any obvious errors:

   kubectl logs -n opentelemetry deployment/opentelemetry-collector
   

Deploy the telemetry generator

You can test your configuration by using the open-source telemetrygen tool. This app generates telemetry and sends it to the collector.

1. To deploy the telemetrygen app in the opentelemetry-demo namespace, run the following command:

   kubectl apply -f https://raw.githubusercontent.com/GoogleCloudPlatform/otlp-k8s-ingest/refs/heads/main/sample/app.yaml
   

2. After you create the deployment, it might take a while for the pods to be created and start running. To check the status of the pods, run the following command:

   kubectl get po -n opentelemetry-demo -w
   

   To stop watching the pod status, enter Ctrl-C to stop the command.

Query a metric by using Metrics Explorer

The telemetrygen tool writes to a metric called gen. You can query this metric from both the query-builder interface and the PromQL query editor in Metrics Explorer.

In the Google Cloud console, go to the  Metrics explorer page:

Go to Metrics explorer

If you use the search bar to find this page, then select the result whose subheading is Monitoring.

• If you use the Metrics Explorer query-builder interface, then the full name of the metric is prometheus.googleapis.com/gen/gauge.
• If you use the PromQL query editor, you can query the metric by using the name gen.

The following image shows a chart of the gen metric in Metrics Explorer:

Delete the cluster

After verifying your deployment by querying the metric, you can delete the cluster. To delete the cluster, run the following command:

gcloud container clusters delete --location CLUSTER_LOCATION otlp-test --project PROJECT_ID

What's next

• For information about using an OpenTelemetry Collector and the Telemetry API with OpenTelemetry zero-code instrumentation, see Use OpenTelemetry zero-code instrumentation for Java.
• For information about sending metrics from applications that use SDKs, see Use SDKs to send metrics from applications.
• For information about migrating to the otlphttp exporter from another exporter, see Migrate to the OTLP exporter.
• To learn more about the Telemetry API, see Telemetry (OTLP) API overview.