×

In OKD 4.17, you can access web service APIs for some monitoring components from the command line interface (CLI).

In certain situations, accessing API endpoints can degrade the performance and scalability of your cluster, especially if you use endpoints to retrieve, send, or query large amounts of metrics data.

To avoid these issues, follow these recommendations:

  • Avoid querying endpoints frequently. Limit queries to a maximum of one every 30 seconds.

  • Do not try to retrieve all metrics data via the /federate endpoint for Prometheus. Query it only when you want to retrieve a limited, aggregated data set. For example, retrieving fewer than 1,000 samples for each request helps minimize the risk of performance degradation.

About accessing monitoring web service APIs

You can directly access web service API endpoints from the command line for the following monitoring stack components:

  • Prometheus

  • Alertmanager

  • Thanos Ruler

  • Thanos Querier

To access Thanos Ruler and Thanos Querier service APIs, the requesting account must have get permission on the namespaces resource, which can be granted by binding the cluster-monitoring-view cluster role to the account.

When you access web service API endpoints for monitoring components, be aware of the following limitations:

  • You can only use bearer token authentication to access API endpoints.

  • You can only access endpoints in the /api path for a route. If you try to access an API endpoint in a web browser, an Application is not available error occurs. To access monitoring features in a web browser, use the OKD web console to review monitoring dashboards.

Additional resources

Accessing a monitoring web service API

The following example shows how to query the service API receivers for the Alertmanager service used in core platform monitoring. You can use a similar method to access the prometheus-k8s service for core platform Prometheus and the thanos-ruler service for Thanos Ruler.

Prerequisites
  • You are logged in to an account that is bound against the monitoring-alertmanager-edit role in the openshift-monitoring namespace.

  • You are logged in to an account that has permission to get the Alertmanager API route.

    If your account does not have permission to get the Alertmanager API route, a cluster administrator can provide the URL for the route.

Procedure
  1. Extract an authentication token by running the following command:

    $ TOKEN=$(oc whoami -t)
  2. Extract the alertmanager-main API route URL by running the following command:

    $ HOST=$(oc -n openshift-monitoring get route alertmanager-main -ojsonpath={.status.ingress[].host})
  3. Query the service API receivers for Alertmanager by running the following command:

    $ curl -H "Authorization: Bearer $TOKEN" -k "https://$HOST/api/v2/receivers"

Querying metrics by using the federation endpoint for Prometheus

You can use the federation endpoint for Prometheus to scrape platform and user-defined metrics from a network location outside the cluster. To do so, access the Prometheus /federate endpoint for the cluster via an OKD route.

A delay in retrieving metrics data occurs when you use federation. This delay can affect the accuracy and timeliness of the scraped metrics.

Using the federation endpoint can also degrade the performance and scalability of your cluster, especially if you use the federation endpoint to retrieve large amounts of metrics data. To avoid these issues, follow these recommendations:

  • Do not try to retrieve all metrics data via the federation endpoint for Prometheus. Query it only when you want to retrieve a limited, aggregated data set. For example, retrieving fewer than 1,000 samples for each request helps minimize the risk of performance degradation.

  • Avoid frequent querying of the federation endpoint for Prometheus. Limit queries to a maximum of one every 30 seconds.

If you need to forward large amounts of data outside the cluster, use remote write instead. For more information, see the Configuring remote write storage section.

Prerequisites
  • You have installed the OpenShift CLI (oc).

  • You have access to the cluster as a user with the cluster-monitoring-view cluster role or have obtained a bearer token with get permission on the namespaces resource.

    You can only use bearer token authentication to access the Prometheus federation endpoint.

  • You are logged in to an account that has permission to get the Prometheus federation route.

    If your account does not have permission to get the Prometheus federation route, a cluster administrator can provide the URL for the route.

Procedure
  1. Retrieve the bearer token by running the following the command:

    $ TOKEN=$(oc whoami -t)
  2. Get the Prometheus federation route URL by running the following command:

    $ HOST=$(oc -n openshift-monitoring get route prometheus-k8s-federate -ojsonpath={.status.ingress[].host})
  3. Query metrics from the /federate route. The following example command queries up metrics:

    $ curl -G -k -H "Authorization: Bearer $TOKEN" https://$HOST/federate --data-urlencode 'match[]=up'
    Example output
    # TYPE up untyped
    up{apiserver="kube-apiserver",endpoint="https",instance="10.0.143.148:6443",job="apiserver",namespace="default",service="kubernetes",prometheus="openshift-monitoring/k8s",prometheus_replica="prometheus-k8s-0"} 1 1657035322214
    up{apiserver="kube-apiserver",endpoint="https",instance="10.0.148.166:6443",job="apiserver",namespace="default",service="kubernetes",prometheus="openshift-monitoring/k8s",prometheus_replica="prometheus-k8s-0"} 1 1657035338597
    up{apiserver="kube-apiserver",endpoint="https",instance="10.0.173.16:6443",job="apiserver",namespace="default",service="kubernetes",prometheus="openshift-monitoring/k8s",prometheus_replica="prometheus-k8s-0"} 1 1657035343834
    ...

Accessing metrics from outside the cluster for custom applications

You can query Prometheus metrics from outside the cluster when monitoring your own services with user-defined projects. Access this data from outside the cluster by using the thanos-querier route.

This access only supports using a bearer token for authentication.

Prerequisites
  • You have deployed your own service, following the "Enabling monitoring for user-defined projects" procedure.

  • You are logged in to an account with the cluster-monitoring-view cluster role, which provides permission to access the Thanos Querier API.

  • You are logged in to an account that has permission to get the Thanos Querier API route.

    If your account does not have permission to get the Thanos Querier API route, a cluster administrator can provide the URL for the route.

Procedure
  1. Extract an authentication token to connect to Prometheus by running the following command:

    $ TOKEN=$(oc whoami -t)
  2. Extract the thanos-querier API route URL by running the following command:

    $ HOST=$(oc -n openshift-monitoring get route thanos-querier -ojsonpath={.status.ingress[].host})
  3. Set the namespace to the namespace in which your service is running by using the following command:

    $ NAMESPACE=ns1
  4. Query the metrics of your own services in the command line by running the following command:

    $ curl -H "Authorization: Bearer $TOKEN" -k "https://$HOST/api/v1/query?" --data-urlencode "query=up{namespace='$NAMESPACE'}"

    The output shows the status for each application pod that Prometheus is scraping:

    The formatted example output
    {
      "status": "success",
      "data": {
        "resultType": "vector",
        "result": [
          {
            "metric": {
              "__name__": "up",
              "endpoint": "web",
              "instance": "10.129.0.46:8080",
              "job": "prometheus-example-app",
              "namespace": "ns1",
              "pod": "prometheus-example-app-68d47c4fb6-jztp2",
              "service": "prometheus-example-app"
            },
            "value": [
              1591881154.748,
              "1"
            ]
          }
        ],
      }
    }

    The formatted example output uses a filtering tool, such as jq, to provide the formatted indented JSON. See the jq Manual for more information about using jq.

Resources reference for the Cluster Monitoring Operator

This document describes the following resources deployed and managed by the Cluster Monitoring Operator (CMO):

Use this information when you want to configure API endpoint connections to retrieve, send, or query metrics data.

In certain situations, accessing API endpoints can degrade the performance and scalability of your cluster, especially if you use endpoints to retrieve, send, or query large amounts of metrics data.

To avoid these issues, follow these recommendations:

  • Avoid querying endpoints frequently. Limit queries to a maximum of one every 30 seconds.

  • Do not try to retrieve all metrics data via the /federate endpoint for Prometheus. Query it only when you want to retrieve a limited, aggregated data set. For example, retrieving fewer than 1,000 samples for each request helps minimize the risk of performance degradation.

CMO routes resources

openshift-monitoring/alertmanager-main

Expose the /api endpoints of the alertmanager-main service via a router.

openshift-monitoring/prometheus-k8s

Expose the /api endpoints of the prometheus-k8s service via a router.

openshift-monitoring/prometheus-k8s-federate

Expose the /federate endpoint of the prometheus-k8s service via a router.

openshift-user-workload-monitoring/federate

Expose the /federate endpoint of the prometheus-user-workload service via a router.

openshift-monitoring/thanos-querier

Expose the /api endpoints of the thanos-querier service via a router.

openshift-user-workload-monitoring/thanos-ruler

Expose the /api endpoints of the thanos-ruler service via a router.

CMO services resources

openshift-monitoring/prometheus-operator-admission-webhook

Expose the admission webhook service which validates PrometheusRules and AlertmanagerConfig custom resources on port 8443.

openshift-user-workload-monitoring/alertmanager-user-workload

Expose the user-defined Alertmanager web server within the cluster on the following ports:

  • Port 9095 provides access to the Alertmanager endpoints. Granting access requires binding a user to the monitoring-alertmanager-api-reader role (for read-only operations) or monitoring-alertmanager-api-writer role in the openshift-user-workload-monitoring project.

  • Port 9092 provides access to the Alertmanager endpoints restricted to a given project. Granting access requires binding a user to the monitoring-rules-edit cluster role or monitoring-edit cluster role in the project.

  • Port 9097 provides access to the /metrics endpoint only. This port is for internal use, and no other usage is guaranteed.

openshift-monitoring/alertmanager-main

Expose the Alertmanager web server within the cluster on the following ports:

  • Port 9094 provides access to all the Alertmanager endpoints. Granting access requires binding a user to the monitoring-alertmanager-view role (for read-only operations) or the monitoring-alertmanager-edit role in the openshift-monitoring project.

  • Port 9092 provides access to the Alertmanager endpoints restricted to a given project. Granting access requires binding a user to the monitoring-rules-edit cluster role or monitoring-edit cluster role in the project.

  • Port 9097 provides access to the /metrics endpoint only. This port is for internal use, and no other usage is guaranteed.

openshift-monitoring/kube-state-metrics

Expose kube-state-metrics /metrics endpoints within the cluster on the following ports:

  • Port 8443 provides access to the Kubernetes resource metrics. This port is for internal use, and no other usage is guaranteed.

  • Port 9443 provides access to the internal kube-state-metrics metrics. This port is for internal use, and no other usage is guaranteed.

openshift-monitoring/metrics-server

Expose the metrics-server web server on port 443. This port is for internal use, and no other usage is guaranteed.

openshift-monitoring/monitoring-plugin

Expose the monitoring plugin service on port 9443. This port is for internal use, and no other usage is guaranteed.

openshift-monitoring/node-exporter

Expose the /metrics endpoint on port 9100. This port is for internal use, and no other usage is guaranteed.

openshift-monitoring/openshift-state-metrics

Expose openshift-state-metrics /metrics endpoints within the cluster on the following ports:

  • Port 8443 provides access to the OpenShift resource metrics. This port is for internal use, and no other usage is guaranteed.

  • Port 9443 provides access to the internal openshift-state-metrics metrics. This port is for internal use, and no other usage is guaranteed.

openshift-monitoring/prometheus-k8s

Expose the Prometheus web server within the cluster on the following ports:

  • Port 9091 provides access to all the Prometheus endpoints. Granting access requires binding a user to the cluster-monitoring-view cluster role.

  • Port 9092 provides access to the /metrics and /federate endpoints only. This port is for internal use, and no other usage is guaranteed.

openshift-user-workload-monitoring/prometheus-operator

Expose the /metrics endpoint on port 8443. This port is for internal use, and no other usage is guaranteed.

openshift-monitoring/prometheus-operator

Expose the /metrics endpoint on port 8443. This port is for internal use, and no other usage is guaranteed.

openshift-user-workload-monitoring/prometheus-user-workload

Expose the Prometheus web server within the cluster on the following ports:

  • Port 9091 provides access to the /metrics endpoint only. This port is for internal use, and no other usage is guaranteed.

  • Port 9092 provides access to the /federate endpoint only. Granting access requires binding a user to the cluster-monitoring-view cluster role.

This also exposes the /metrics endpoint of the Thanos sidecar web server on port 10902. This port is for internal use, and no other usage is guaranteed.

openshift-monitoring/telemeter-client

Expose the /metrics endpoint on port 8443. This port is for internal use, and no other usage is guaranteed.

openshift-monitoring/thanos-querier

Expose the Thanos Querier web server within the cluster on the following ports:

  • Port 9091 provides access to all the Thanos Querier endpoints. Granting access requires binding a user to the cluster-monitoring-view cluster role.

  • Port 9092 provides access to the /api/v1/query, /api/v1/query_range/, /api/v1/labels, /api/v1/label/*/values, and /api/v1/series endpoints restricted to a given project. Granting access requires binding a user to the view cluster role in the project.

  • Port 9093 provides access to the /api/v1/alerts, and /api/v1/rules endpoints restricted to a given project. Granting access requires binding a user to the monitoring-rules-edit cluster role, monitoring-edit cluster role or monitoring-rules-view cluster role in the project.

  • Port 9094 provides access to the /metrics endpoint only. This port is for internal use, and no other usage is guaranteed.

openshift-user-workload-monitoring/thanos-ruler

Expose the Thanos Ruler web server within the cluster on the following ports:

  • Port 9091 provides access to all Thanos Ruler endpoints. Granting access requires binding a user to the cluster-monitoring-view cluster role.

  • Port 9092 provides access to the /metrics endpoint only. This port is for internal use, and no other usage is guaranteed.

This also exposes the gRPC endpoints on port 10901. This port is for internal use, and no other usage is guaranteed.

openshift-monitoring/cluster-monitoring-operator

Expose the /metrics endpoint on port 8443. This port is for internal use, and no other usage is guaranteed.