Apache Flink

This document describes how to configure your Google Kubernetes Engine deployment so that you can use Google Cloud Managed Service for Prometheus to collect metrics from the Flink exporter. This document shows you how to do the following:

Set up the Flink exporter to report metrics.
Configure a PodMonitoring resource for Managed Service for Prometheus to collect the exported metrics.
Access a dashboard in Cloud Monitoring to view the metrics.
Configure alerting rules to monitor the metrics.

These instructions apply only if you are using managed collection with Managed Service for Prometheus. If instead you are using self-deployed collection, then see the Flink documentation for installation information.

For information about Flink, see Apache Flink.

Prerequisites

To collect metrics from the Flink exporter by using Managed Service for Prometheus and managed collection, your deployment must meet the following requirements:

Your cluster must be running Google Kubernetes Engine version 1.21.4-gke.300 or later.
You must be running Managed Service for Prometheus with managed collection enabled. For more information, see Get started with managed collection.
To use dashboards available in Cloud Monitoring for the Flink integration, you must use flink version 1.17 or later.
For more information about available dashboards, see View dashboards.

Flink exposes Prometheus-format metrics when configured with metrics.reporter.prom.factory.class: org.apache.flink.metrics.prometheus.PrometheusReporterFactory.

If you deployed Flink with the official getting started manifests, add this new option to the ConfigMap:

  apiVersion: v1
  kind: ConfigMap
  metadata:
    name: flink-config
    labels:
      app: flink
  data:
    flink-conf.yaml: |+
      ...
  +   metrics.reporter.prom.factory.class: org.apache.flink.metrics.prometheus.PrometheusReporterFactory

If you deployed Flink with the official operator, add this new option to the FlinkDeployment's spec.flinkConfiguration field:

  apiVersion: flink.apache.org/v1beta1
  kind: FlinkDeployment
  metadata:
    name: basic-example
  spec:
    image: flink:1.17
    flinkVersion: v1_17
    flinkConfiguration:
      taskmanager.numberOfTaskSlots: "2"
  +   metrics.reporter.prom.factory.class: org.apache.flink.metrics.prometheus.PrometheusReporterFactory

Alternatively, you can specify the Prometheus reporter as a default option within the Flink operator configuration.

To verify that the Flink exporter is emitting metrics on the expected endpoints, do the following:

Set up port forwarding by using the following command:

kubectl -n NAMESPACE_NAME port-forward POD_NAME 9249

Access the endpoint localhost:9249/metrics by using the browser or the curl utility in another terminal session.

Define a PodMonitoring resource

For target discovery, the Managed Service for Prometheus Operator requires a PodMonitoring resource that corresponds to the Flink exporter in the same namespace.

You can use the following PodMonitoring configuration:

View on GitHub

apiVersion: monitoring.googleapis.com/v1
kind: PodMonitoring
metadata:
  name: flink
  labels:
    app.kubernetes.io/name: flink
    app.kubernetes.io/part-of: google-cloud-managed-prometheus
spec:
  endpoints:
  - port: 9249
    scheme: http
    interval: 30s
    path: /metrics
  selector:
    matchLabels:
      app.kubernetes.io/name: flink

Ensure that the values of the port and matchLabels fields match those of the Flink pods you want to monitor. Flink exposes metrics on port 9249 by default when metrics are enabled.

To apply configuration changes from a local file, run the following command:

kubectl apply -n NAMESPACE_NAME -f FILE_NAME

You can also use Terraform to manage your configurations.

Define rules and alerts

You can use the following Rules configuration to define alerts on your Flink metrics:

View on GitHub

apiVersion: monitoring.googleapis.com/v1
kind: Rules
metadata:
  name: flink-rules
  labels:
    app.kubernetes.io/component: rules
    app.kubernetes.io/name: flink-rules
    app.kubernetes.io/part-of: google-cloud-managed-prometheus
spec:
  groups:
  - name: flink
    interval: 30s
    rules:
    - alert: FlinkHighJVMMemoryNonHeapUsage
      annotations:
        description: |-
          Flink high jvm memory non-heap usage
            VALUE = {{ $value }}
            LABELS: {{ $labels }}
        summary: Flink high jvm memory non-heap usage (instance {{ $labels.instance }})
      expr: (flink_jobmanager_Status_JVM_Memory_NonHeap_Used/flink_jobmanager_Status_JVM_Memory_NonHeap_Max) >= 0.9
      for: 5m
      labels:
        severity: critical
    - alert: FlinkHighJVMMemoryHeapUsage
      annotations:
        description: |-
          Flink high jvm memory heap usage
            VALUE = {{ $value }}
            LABELS: {{ $labels }}
        summary: Flink high jvm memory heap usage (instance {{ $labels.instance }})
      expr: (flink_jobmanager_Status_JVM_Memory_Heap_Used/flink_jobmanager_Status_JVM_Memory_Heap_Max) >= 0.9
      for: 5m
      labels:
        severity: warning

To apply configuration changes from a local file, run the following command:

kubectl apply -n NAMESPACE_NAME -f FILE_NAME

You can also use Terraform to manage your configurations.

For more information about applying rules to your cluster, see Managed rule evaluation and alerting.

You can adjust the alert thresholds to suit your application.

Verify the configuration

You can use Metrics Explorer to verify that you correctly configured the Flink exporter. It might take one or two minutes for Cloud Monitoring to ingest your metrics.

To verify the metrics are ingested, do the following:

In the navigation panel of the Google Cloud console, select Monitoring, and then select Metrics explorer:
Go to Metrics explorer
In the toolbar of the Select a metric pane, select the button whose name starts with Code.
Select PromQL in the Language toggle. The language toggle is in the same toolbar that lets you format your query.

Enter and run the following query:

up{job="flink", cluster="CLUSTER_NAME", namespace="NAMESPACE_NAME"}

View dashboards

The Cloud Monitoring integration includes "Prometheus" dashboards. Dashboards are automatically installed when you configure the integration. You can also view static previews of dashboards without installing the integration.

To view an installed dashboard, do the following:

In the navigation panel of the Google Cloud console, select Monitoring, and then select Dashboards:
Go to Dashboards
Select the Dashboard List tab.
Choose the Integrations category.
Click the name of the dashboard you want to view.

To view a static preview of the dashboard, do the following:

In the navigation panel of the Google Cloud console, select Monitoring, and then select Integrations:
Go to Integrations
Click the Kubernetes Engine deployment-platform filter.
Locate the Apache Flink integration and click View Details.
Select the Dashboards tab.

Troubleshooting

For information about troubleshooting metric ingestion problems, see Problems with collection from exporters in Troubleshooting ingestion-side problems.