kibana-prometheus-exporter

Prometheus Exporter for Kibana (7.*., 8.*)

1. Usage
   1. Docker
   2. Kubernetes
      1. Helm Chart
2. Metrics
3. TODO
4. Contributing
5. License

A standalone Prometheus exporter for Kibana metrics inspired by the Kibana Prometheus Exporter Plugin.

This makes use of the /api/status endpoint to gather and convert metrics to the Prometheus OpenMetrics format.

Usage

NOTE: Currently tested against below Kibana versions only. Checkout releases and Docker images for support for other Kibana versions.

1. 8.5
2. 8.7

Please open an issue if you see errors or missing metrics with the Kibana version you're using.

Match the Kibana version with the release tag (ex: release v7.5.x.2 will work with Kibana 7.5.x versions. It's possible it will continue to work for a few more minor releases, but this depends on what Elastic decides to do with the idea of semantic versioning)

First 3 sections of the release tag represents the Kibana version compatibility, while the last section indicates patching increments. Older releases will work with the corresponding Kibana versions, however some newer features such as the /healthz endpoint will not be backported to them, at least initially. Please open an issue if you are working with an older Kibana version but still needs them updated.

# expose metrics from the local Kibana instance using the provided username and password
kibana-exporter -kibana.uri http://localhost:5601 -kibana.username elastic -kibana.password password

By default, the Exporter exposes the /metrics endpoint at port 9684. If needed this port (and the endpoint) can be overridden.

# expose the /metrics endpoint at port 8080
kibana-exporter -kibana.uri http://localhost:5601 -web.listen-address :8080

# expose metrics using /scrape endpint
kibana-exporter -kibana.uri http://localhost:5601 -web.telemetry-path "/scrape"

# skip TLS verification for self-signed Kibana certificates
kibana-exporter -kibana.uri https://kibana.local:5601 -kibana.skip-tls true

Flags

  -debug
        Output verbose details during metrics collection, use for development only
  -kibana.password string
        The password to use for Kibana API
  -kibana.skip-tls
        Skip TLS verification for TLS secured Kibana URLs
  -kibana.uri string
        The Kibana API to fetch metrics from
  -kibana.username string
        The username to use for Kibana API
  -wait
        Wait for Kibana to be responsive before starting, setting this to false would cause the exporter to error out instead of waiting
  -web.listen-address string
        The address to listen on for HTTP requests. (default ":9684")
  -web.telemetry-path string
        The address to listen on for HTTP requests. (default "/metrics")

Docker

The Docker Image chamilad/kibana-prometheus-exporter can be used directly to run the exporter in a Dockerized environment. The Container filesystem only contains the statically linked binary, so that it can be run independently.

docker run -p 9684:9684 -it chamilad/kibana-prometheus-exporter:v7.10.x.2 -kibana.username elastic -kibana.password password -kibana.uri https://elasticcloud.kibana.aws.found.io

Refer to the Makefile and the Dockerfile for more details.

Kubernetes

Refer the artifacts in k8s directory. There is a Deployment and a Service that exposes the Deployment.

kubectl apply -f k8s/kibana-prometheus-exporter.yaml

$  kubectl get all -l app=kibana-prometheus-exporter
  NAME                                             READY   STATUS    RESTARTS   AGE
  pod/kibana-prometheus-exporter-b8c888bcd-66kvx   1/1     Running   0          16s

  NAME                                 TYPE        CLUSTER-IP     EXTERNAL-IP   PORT(S)    AGE
  service/kibana-prometheus-exporter   ClusterIP   10.96.252.18   <none>        9684/TCP   16s

  NAME                                         READY   UP-TO-DATE   AVAILABLE   AGE
  deployment.apps/kibana-prometheus-exporter   1/1     1            1           16s

  NAME                                                   DESIRED   CURRENT   READY   AGE
  replicaset.apps/kibana-prometheus-exporter-b8c888bcd   1         1         1       16s

With these artifacts deployed, the following Prometheus scrape configuration can be used to scrape the metrics.

- job_name: "kibana"
  scrape_interval: 1m
  metrics_path: "/metrics"
  kubernetes_sd_configs:
    - role: service
  relabel_configs:
    - source_labels: [__meta_kubernetes_service_label_app]
      regex: "kibana-exporter"
      action: keep
    - source_labels: [__meta_kubernetes_namespace]
      action: replace
      target_label: kubernetes_namespace
    - source_labels:
        [__address__, __meta_kubernetes_service_annotation_prometheus_io_port]
      target_label: __address__
      regex: ([^:]+)(?::\d+)?;(\d+)
      replacement: $1:$2

1. The scrape_interval for the job is kept to once per minute. This is to reduce the load on the ElasticSearch cluster, by frequent API calls.
2. The port to connect is detected through a K8s Service annotation, prometheus.io/port.
3. The metrics will end up with the label job: kibana

/healthz Endpoint

A simple GET /healthz endpoint has been provided to be used for health checks and liveness probes in environments like K8s.

Helm Chart

@pavdmyt has developed a Helm chart that can be successfully used to deploy the Kibana Exporter on a K8s cluster available at AnchorFree. To use the chart, simply add the Helm repository and install the chart as follows.

$ helm repo add afcharts https://anchorfree.github.io/helm-charts
$ helm repo update
$ helm install <release-name> afcharts/kibana-exporter

Metrics

The metrics exposed by this Exporter are the following.

Grafana Dashboard

A simple starter dashboard json file is included in the repository for Grafana. This can be used as a starting place to customise for own needs.

Binary and Image Verification

The binary and the Docker images are signed with Sigstore Cosign. The public key is available in the release artefacts as cosign-<release_tag>.pub.

Install cosign and use the following commands on Linux to verify the binaries and the images.

Replace <release_tag>, <binary>, and <image> values as necessary. Each release note will have release specific commands.

# download the binary, signature, and the public key
wget https://github.com/chamilad/kibana-prometheus-exporter/releases/download/<release_tag>/<binary>
wget https://github.com/chamilad/kibana-prometheus-exporter/releases/download/<release_tag>/<binary>.sig
wget https://github.com/chamilad/kibana-prometheus-exporter/releases/download/<release_tag>/cosign-<release_tag>.pub

# verify the images and the binary
cosign verify --key cosign-<release_tag>.pub chamilad/<image>:<release_tag>
cosign verify --key cosign-<release_tag>.pub ghcr.io/chamilad/<image>:<release_tag>
cosign verify-blob --key cosign-<release_tag>.pub --signature <binary>.sig <binary>

Additionally, a Software Bill of Materials (SBOM), in both SPDX and JSON formats are attached to the artefacts in each release.

TODO

1. Test other versions and edge cases more
2. Come up with a way to keep up with Kibana API changes
3. Add more metrics related to the scrape job itself
4. Add a Grafana dashboards with (Prometheus) alerts

Contributing

More metrics, useful tweaks, samples, bug fixes, and any other form of contributions are welcome. Please fork, modify, and open a PR. Please open a GitHub issue for observed bugs or feature requests. I will try to attend to them when possible.

I haven't figured out testing unfortunately, once I have it'll be easier to test changes. Until then, start up basic Elastic and Kibana instances if you're testing dev changes.

License

The contents of this repository are licensed under Apache V2 License.