TfL Cycles Exporter

Prometheus exporter for Transport for London cycle hire availability data.

Powered by TfL Open Data.

Metrics

At the time of writing, there are over 700 docking stations. The exporter will expose 4 time series for each one:

# HELP tflcycles_bicycles_available The number of in-service, conventional bikes available for hire.
# TYPE tflcycles_bicycles_available gauge
tflcycles_bicycles_available{station="Stonecutter Street, Holborn"} 2
...
# HELP tflcycles_docks The total number of docks at the station, including those that are out of service.
# TYPE tflcycles_docks gauge
tflcycles_docks{station="Stonecutter Street, Holborn"} 21
...
# HELP tflcycles_docks_available The number of in-service, vacant docks to which a bike can be returned.
# TYPE tflcycles_docks_available gauge
tflcycles_docks_available{station="Stonecutter Street, Holborn"} 19
...
# HELP tflcycles_ebikes_available The number of in-service e-bikes available for hire.
# TYPE tflcycles_ebikes_available gauge
tflcycles_ebikes_available{station="Stonecutter Street, Holborn"} 0
...

Configuration

Download the latest release for your platform, extract, and invoke:

$ curl -LO https://github.com/gebn/tflcycles_exporter/releases/download/v1.1.2/tflcycles_exporter-1.1.2.linux-amd64.tar.gz
$ tar xf tflcycles_exporter-1.1.2.linux-amd64.tar.gz
$ cd tflcycles_exporter-1.1.2.linux-amd64
$ ./tflcycles_exporter

By default, the exporter will listen on port 9722. Visit http://localhost:9722/stations to see the metrics.

Rate Limits

The exporter uses TfL's BikePoint API to retrieve docking station information. Registration is optional, however it provides success and latency metrics about your API requests. An application key can be passed in an APP_KEY environment variable when starting the exporter, and it will be used automatically.

Systemd

The following steps will be suitable for the majority of Linux users.

1. Extract the release archive to /opt/tflcycles_exporter.

2. Copy tflcycles_exporter.service into /etc/systemd/system, and open the file in your favourite editor.

   1. Ensure User is set to a suitable value. A dedicated account could be created with:

      # useradd -s /usr/sbin/nologin -r -M tflcycles_exporter
      

   2. Set APP_KEY if desired.

3. Execute the following as root:

   systemctl enable tflcycles_exporter.service  # start on boot
   systemctl start tflcycles_exporter.service   # launch exporter
   systemctl status tflcycles_exporter.service  # check running smoothly
   

Container

Images are published to Docker Hub each push. In a Kubernetes context, the / endpoint is efficient to serve, so is suitable for liveness and readiness probes.

Prometheus

The exporter exposes its own direct-instrumentation metrics at /metrics, which can be scraped normally.

TfL data is exposed at /stations. Each request to this endpoint will trigger a call to the BikePoint API and render the results as metrics. The source data is updated at most once per minute, so the scrape interval should not be below 1m. Setting it below this will still work, however it will needlessly re-retrieve the same values from TfL's API, and use up your request limit unnecessarily.

scrape_configs:

- job_name: tflcycles-exporter
  static_configs:
  - targets:
    - localhost:9722

- job_name: tflcycles
  scrape_interval: 1m
  metrics_path: /stations
  static_configs:
  - targets:
    - localhost:9722

If Prometheus is deployed with multiple replicas, and you plan to colocate an exporter instance next to each one, the /metrics job should not be deduplicated, as these are separate processes. You may need to use a hostname other than localhost to ensure distinct label sets. The /stations job can be deduplicated safely, as all exporters should return the same thing within a given minute.