Skip to content

Use Kafka as a remote storage database for Prometheus (remote write only)

License

Notifications You must be signed in to change notification settings

Telefonica/prometheus-kafka-adapter

Repository files navigation

prometheus-kafka-adapter

CI

Prometheus-kafka-adapter is a service which receives Prometheus metrics through remote_write, marshal into JSON and sends them into Kafka.

output

It is able to write JSON or Avro-JSON messages in a kafka topic, depending on the SERIALIZATION_FORMAT configuration variable.

JSON

{
  "timestamp": "1970-01-01T00:00:00Z",
  "value": "9876543210",
  "name": "up",

  "labels": {
    "__name__": "up",
    "label1": "value1",
    "label2": "value2"
  }
}

timestamp and value are reserved values, and can't be used as label names. __name__ is a special label that defines the name of the metric and is copied as name to the top level for convenience.

Avro JSON

The Avro-JSON serialization is the same. See the Avro schema.

configuration

prometheus-kafka-adapter

There is a docker image telefonica/prometheus-kafka-adapter:1.9.1 available on Docker Hub.

Prometheus-kafka-adapter listens for metrics coming from Prometheus and sends them to Kafka. This behaviour can be configured with the following environment variables:

  • KAFKA_BROKER_LIST: defines kafka endpoint and port, defaults to kafka:9092.
  • KAFKA_TOPIC: defines kafka topic to be used, defaults to metrics. Could use go template, labels are passed (as a map) to the template: e.g: metrics.{{ index . "__name__" }} to use per-metric topic. Two template functions are available: replace ({{ index . "__name__" | replace "message" "msg" }}) and substring ({{ index . "__name__" | substring 0 5 }})
  • KAFKA_COMPRESSION: defines the compression type to be used, defaults to none.
  • KAFKA_BATCH_NUM_MESSAGES: defines the number of messages to batch write, defaults to 10000.
  • SERIALIZATION_FORMAT: defines the serialization format, can be json, avro-json, defaults to json.
  • PORT: defines http port to listen, defaults to 8080, used directly by gin.
  • BASIC_AUTH_USERNAME: basic auth username to be used for receive endpoint, defaults is no basic auth.
  • BASIC_AUTH_PASSWORD: basic auth password to be used for receive endpoint, defaults is no basic auth.
  • LOG_LEVEL: defines log level for logrus, can be debug, info, warn, error, fatal or panic, defaults to info.
  • GIN_MODE: manage gin debug logging, can be debug or release.
  • MATCH: filter which metrics and labels to send to kafka. Used to limit the amount of data that is sent to kafka - by default everything is send. Example: "['metric-1{label-1="label-value-1"}','metric-2']"

To connect to Kafka over SSL define the following additional environment variables:

  • KAFKA_SSL_CLIENT_CERT_FILE: Kafka SSL client certificate file, defaults to ""
  • KAFKA_SSL_CLIENT_KEY_FILE: Kafka SSL client certificate key file, defaults to ""
  • KAFKA_SSL_CLIENT_KEY_PASS: Kafka SSL client certificate key password (optional), defaults to ""
  • KAFKA_SSL_CA_CERT_FILE: Kafka SSL broker CA certificate file, defaults to ""

To connect to Kafka over SASL/SCRAM authentication define the following additional environment variables:

  • KAFKA_SECURITY_PROTOCOL: Kafka client used protocol to communicate with brokers, must be set if SASL is going to be used, either plain or with SSL
  • KAFKA_SASL_MECHANISM: SASL mechanism to use for authentication, defaults to ""
  • KAFKA_SASL_USERNAME: SASL username for use with the PLAIN and SASL-SCRAM-.. mechanisms, defaults to ""
  • KAFKA_SASL_PASSWORD: SASL password for use with the PLAIN and SASL-SCRAM-.. mechanism, defaults to ""
  • KAFKA_SSL_CA_CERT_FILE: Kafka SSL broker CA certificate file, defaults to ""

When deployed in a Kubernetes cluster using Helm and using a Kafka external to the cluster, it might be necessary to define the kafka hostname resolution locally (this fills the /etc/hosts of the container). Use a custom values.yaml file with section hostAliases (as mentioned in default values.yaml).

prometheus

Prometheus needs to have a remote_write url configured, pointing to the '/receive' endpoint of the host and port where the prometheus-kafka-adapter service is running. For example:

remote_write:
  - url: "http://prometheus-kafka-adapter:8080/receive"

When deployed in a Kubernetes cluster using Helm and using an external Prometheus, it might be necessary to expose prometheus-kafka-adapter input port as a node port. Use a custom values.yaml file to set service.type: NodePort and service.nodeport: <PortNumber> (see comments in default values.yaml)

development

The provided Makefile can do basic linting/building for you simply:

  • make fmt -> basic formatting.
  • make test -> runs go test fixtures
  • make vet -> runs go vet against the package
  • make vendor-update -> ensure dependencies are up to date
  • make -> builds libc and musl based binaries, including testing and vetting code

contributing

With issues:

  • Use the search tool before opening a new issue.
  • Please provide source code and commit sha if you find a bug.
  • Review existing issues and provide feedback or react to them.

With pull requests:

  • Open your pull request against master
  • It should pass all tests in the continuous integration pipeline (Github Actions).
  • You should add/modify tests to cover your proposed code changes.
  • If your pull request contains a new feature, please document it in this README.

license

Copyright 2018 Telefónica

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.