Prometheus-operator comes with pre-configured scrape targets to keep an eye on kubernetes cluster standard components. At some point, you might want to add some custom targets to monitor your application. This page shows you how to achieve it.

• Create a yaml manifest that includes all the extra custom scrape targets you want to add:
  • Sample file: prometheus-additional.yaml

- job_name: "your_custom_job_name"
  static_configs:
  - targets: ["your_endpoint_providing_metrics:your_port"]
  metrics_path: "/a/b/c/metrics/application"

Target configuration settings

Value of “targets” can only be a hostname or ip address (typically: Your application pod’s Service name, e.g. podname.namespace.svc.cluster.local) and the corresponding port.

By default, in case you do NOT specify the “metrics_path”, prometheus will contact http://hostname:port/metrics

In case your application provides metrics to a different path, you must provide it as value of “metrics_path”.

• Create a Secret yaml manifest with name = additional-scrape-configs reading the content from file prometheus-additional.yaml created on step 1 above:

# kubectl create secret generic additional-scrape-configs --from-file=prometheus-additional.yaml --dry-run=client -o yaml > additional-scrape-configs.yaml

• Create a Secret using the yaml manifest generated on step 2 above and make sure to assign it to the same namespace in use by prometheus:

# kubectl apply -f additional-scrape-configs.yaml -n monitoring

• Edit your Prometheus CRD (Custom Resource Definition) and add a reference to your additional scrape configs (new block: spec.additionalScrapeConfigs):

# kubectl edit prometheus/prometheus-kube-prometheus-prometheus -n monitoring

apiVersion: monitoring.coreos.com/v1
kind: Prometheus
metadata:
  annotations:
    meta.helm.sh/release-name: prometheus
    meta.helm.sh/release-namespace: monitoring
  creationTimestamp: "2022-09-15T07:20:00Z"
  generation: 2
  labels:
    app: kube-prometheus-stack-prometheus
    app.kubernetes.io/instance: prometheus
    app.kubernetes.io/managed-by: Helm
    app.kubernetes.io/part-of: kube-prometheus-stack
    app.kubernetes.io/version: 40.0.0
    chart: kube-prometheus-stack-40.0.0
    heritage: Helm
    release: prometheus
  name: prometheus-kube-prometheus-prometheus
  namespace: monitoring
  resourceVersion: "11481588"
  uid: 465362f4-a309-4022-94fb-62f5e22f4828
spec:
  additionalScrapeConfigs:
    key: prometheus-additional.yaml
    name: additional-scrape-configs
. . .

• Restart kube-prometheus and kube-operator pods:

# kubectl delete -n monitoring $(kubectl get pods -o=name -n monitoring -l app=kube-prometheus-stack-operator)
# kubectl delete -n monitoring $(kubectl get pods -o=name -n monitoring -l app.kubernetes.io/instance=prometheus-kube-prometheus-prometheus)

As soon as the new pods come up, metrics collected from your new targets will be accessible from Prometheus/Grafana.

Learn how to add your custom rules to prometheus-operator running on kubernetes

Prometheus-operator comes with a set of pre-installed alerting rules, but typically you want to add extra ones, looking after your own application running on the kubernetes cluster.

This page shows you how.

• Create a new ConfigMap that includes all your new custom groups and related rules. Sample yaml manifest:

apiVersion: monitoring.coreos.com/v1
kind: PrometheusRule
metadata:
  labels:
    prometheus: prometheus-kube-prometheus-prometheus
    role: alert-rules
    release: prometheus
  name: prometheus-test-rules
  namespace: monitoring
spec:
  groups:
  - name: ./test.rules
    rules:
    - alert: No connections to external systems
      expr: (application_test_ext_system_total > 0) and (application_test_ext_system_connections_count_total == 0)
      for: 3m
      labels:
        severity: warning
  - name: ./test-ddos.rules
    rules:
    - alert: Detect anomalies on network traffic
      expr: (( rate(node_network_receive_packets_total[5m]) / rate(node_network_receive_packets_total[5m] offset 5m)) > 10) and ( rate(node_network_receive_packets_total[5m]) > 1000)
      for: 3m
      labels:
        severity: warning

• Restart your prometheus master pod so that the new ConfigMap created above is loaded:

$ kubectl delete -n monitoring $(k get pods -n monitoring -o=name -l operator.prometheus.io/name=prometheus-kube-prometheus-prometheus)

Checking rules validity

Alerting rules are going through a syntax check at prometheus core startup.

This can also be done upfront, using the “promtool” utility, part of prometheus core container.

Syntax:

$ promtool check config your_custom_rules_file.yaml

Sample rules yaml file:

groups:
- name: ./test.rules
  rules:
  - alert: No connections to external systems
    expr: (application_test_ext_system_total > 0) and (application_test_ext_system_connections_count_total == 0)
    for: 3m
    labels:
      severity: warning
- name: ./test-ddos.rules
  rules:
  - alert: Detect anomalies on network traffic
    expr: (( rate(node_network_receive_packets_total[5m]) / rate(node_network_receive_packets_total[5m] offset 5m)) > 10) and ( rate(node_network_receive_packets_total[5m]) > 1000)
    for: 3m
    labels:
      severity: warning

Prometheus does NOT check whether the metrics/labels you provide on the custom rules ConfigMap are valid and existing.

To retrieve the list of available metrics, you can create a new panel on Grafana and take advantage of auto-completion or just use the Prometheus web UI.