prometheus-data-generator

Creates arbitrary prometheus metrics.

Why use this?

When creating Grafana dashboards or Prometheus alerts, it is common to make mistakes. You define a threshold that they have to meet, but when modified the next time you may forget those thresholds.

Using this tool, you can create data with the format you want and thus being able to base alerts and graphics on data that resemble reality.

To use this, you'll create a configuration file in which you will define a metric name, description, type and labels and sequences of certain operations.

For example, you'll be able to create a alarm called http_requests with the labels {path=/login/, return_code=200} which will be updated as you wish.

Configuration

There's an example configuration file called config.yml in the root of the repository. It has the next format:

config:
  - name: number_of_fruits
    description: The number of fruits we have.
    type: gauge
    labels: [name, color]
    sequence:
      - time: 5
        eval_time: 5
        values: 0-20
        operation: inc
        labels:
          name: apple
          color: red
      - time: 5
        eval_time: 5
        values: 0-20
        operation: inc
        labels:
          name: apple
          color: green
      - time: 5
        eval_time: 5
        values: 0-5
        operation: dec
        labels:
          name: apple
          color: green
      - time: 5
        eval_time: 5
        value: 3
        operation: inc
        labels:
          name: apple
          color: yellow

The generated metric will be like this:

number_of_fruits{color="red",name="apple"} 14.0
number_of_fruits{color="green",name="apple"} 10.0
number_of_fruits{color="yellow",name="apple"} 4.0

Supported keywords

• name: The metric name. [Type: string] [Required]
• description: The description to be shown as HELP. [Type: string] [Required]
• type: It should be one of the supported metric types, which you can see in the next section. [Type: string] [Required]
• labels: The labels that will be used with the metric. [Type: list of strings] [Optional]
• sequence.eval_time: Number of seconds that the sequence will be running. [Type: int] [Required]
• sequence.interval: The interval of seconds between each operation will be performed. 1 second is a sane number. [Type: int] [Required]
• sequence.value: The value that the operation will apply. It must be a single value. You must choose between value and values. [Type: int] [Optional]
• sequence.values: The range of values that will randomly be choosed and the operation will apply. It must be two values separed by a dash. You must choose between value and values. [Type: string (int-int / float-float)] [Optional]
• sequence.operation: The operation that will be applied. It only will be used with the gauge type, and you can choose between inc, dec or set. [Optional]
• sequence.labels: The labels of the sequence. They must be used if labels are declared. [Optional]

Supported metric types

The ones defined here.

• Counter
• Gauge
• Histogram
• Summary

Manual use

git clone https://github.com/little-angry-clouds/prometheus-data-generator.git
virtualenv -p python3 venv
pip install -r requirements.txt
python prometheus_data_generator/main.py
curl localhost:9000/metrics/

Use in docker

wget https://raw.githubusercontent.com/little-angry-clouds/prometheus-data-generator/master/config.yml
docker run -ti -v `pwd`/config.yml:/config.yml -p 127.0.0.1:9000:9000 \
    littleangryclouds/prometheus-data-generator
curl localhost:9000/metrics/

Use in kubernetes

There's some example manifests in the kubernetes directory. There's defined a service, configmap, deployment (with configmap-reload configured) and a Service Monitor to be used with the prometheus operator.

You may deploy the manifests:

kubectl create namespace prom-data-gen
kubectl -n prom-data-gen apply -f kubernetes/
kubectl -n prom-data-gen port-forward service/prometheus-data-generator 9000:9000
curl localhost:9000/metrics/

You can edit the configmap as you wish and the configmap-reload will eventually reload the configuration without killing the pod.

Generate prometheus alerts unit tests

TODO

Tests

TODO