Docker Hub rate limits exporter for Prometheus

We should track how many pulls left because Docker Hub limits the number of Docker image downloads. It depends on the account type and publisher programs. Read more at Download rate limit article

Rate-limits-exporter expose metrics for certain Docker Hub account(s) you've provided or for Anonymous user (by default without any credentials). The checks are being executed in the background. Calling /metrics path returns cached metrics.

Usage

Multiple installation scenarios are provided.

Docker

Don't forget to pass valid folder into container (read more at Help below):

docker run -it --rm \
  -v /app/rate-limits:/opt/secrets \
  -e APP_SECRETS_DIR=/opt/secrets \
  -e APP_PORT=8080 \
  -e APP_CHECK_INTERVAL=60 \
  -e LOG_LEVEL=DEBUG \
  -e APP_PUT_SOURCE_IP=False \
  bissquit/rate-limits-exporter:latest

Docker-compose

For testing purposes or for quick review you may use Docker Compose:

docker-compose up -d --build

k8s

Deployment example provided in k8s/ folder. Run it with command:

kubectl -n namespace_name apply -f k8s/deployment.yaml

Help

Exporter looks for the files in directory and expects each file is a separate Docker Hub account where filename is a username of Docker Hub; file content is a password (trailing new line will be removed)

You may pass options both via command line arguments or environment variables:

Command line argument	Environment variable	Description
-h, --help	-	show help message
-d, --directory	APP_SECRETS_DIR	Directory with files. The name of file - username of DockerHub, file content - password. (default: /opt/secrets)
-p, --port	APP_PORT	Port to be listened (default: 8080)
-t, --time	APP_CHECK_INTERVAL	Default time range in seconds to perform rate limits check (default: 60)
-	LOG_LEVEL	Log level based on Python logging module. expected values: DEBUG, INFO, WARNING, ERROR, CRITICAL (default: INFO)
-s, --source	APP_PUT_SOURCE_IP	Source ip address from headers will be inserted into labels. Be careful to use this option because many different ip addressess in the label set will produce a lot of new time series in Prometheus TSDB. Read CAUTION in official article METRIC AND LABEL NAMING

Metrics example:

# HELP dockerhub_ratelimit_current Current max limit for DockerHub account (or for ip address if anonymous access)
# TYPE dockerhub_ratelimit_current gauge
dockerhub_ratelimit_current{dockerhub_user="Anonymous"} 100
# HELP dockerhub_ratelimit_remaining Remaining limit for DockerHub account (or for ip address if anonymous access)
# TYPE dockerhub_ratelimit_remaining gauge
dockerhub_ratelimit_remaining{dockerhub_user="Anonymous"} 100
dockerhub_ratelimit_scrape_error{dockerhub_user="Anonymous"} 0