ziplineai/chronon-fetcher repository overview

⁠Chronon Feature Fetching Service

Docker image that allows you to run the Chronon Feature Service.

⁠Launching the Service

To fetch the latest image, run:

docker pull ziplineai/chronon-fetcher:latest

To run the service, you can use the following command:

docker run \
 -p 9000:9000 \
 ziplineai/chronon-fetcher:latest

⁠Additional Configuration & Environment Variables

There's some additional environment variables that the Docker container can be started up with (using the -e "VAR=VALUE" syntax)

⁠Exporting Metrics

The Chronon library and Vert.x⁠ fetcher web service can be collect metrics using the OpenTelemetry⁠ standard. Currently, the service supports two ways of exporting metrics:

1. OTLP over HTTP: Metrics can be exported (push) to an OTLP endpoint using the EXPORTER_OTLP_ENDPOINT environment variable.
2. Prometheus: Metrics can be exposed on a Prometheus endpoint (pull) using the CHRONON_PROMETHEUS_SERVER_PORT and VERTX_PROMETHEUS_SERVER_PORT environment variables.

⁠Service profiling

The service can be profiled using the Google Cloud Profiler⁠ by setting the ENABLE_GCLOUD_PROFILER environment variable to true. The profiles are exported to the cloud profiler under the service name 'chronon-fetcher'.

⁠Service Endpoints

The service exposes the following HTTP endpoints:

⁠Example Usage

Some example curl calls to the service are shown below. The fetch join and GroupBy endpoints support bulk lookups, so you can pass a list of entity keys to fetch features for multiple entities in a single request.

$ curl http://localhost:9000/v1/joins                                                            
{"joinNames":["search.ranking_listing","search.recommendation_v1"]}

$ curl http://localhost:9000/v1/join/search.ranking_listing/schema | jq
{
  "joinName": "search.ranking_listing",
  "keySchema": "...",
  "valueSchema": "...",
  "schemaHash": "c72b6d",
  ...
}

$ curl -X POST http://localhost:9000/v1/fetch/join/ranking_listing -H 'Content-Type: application/json' -d '[{"listing_id":"12345"}]' | jq
{
  "results": [
    {
      "status": "Success",
      "entityKeys": {
        "listing_id": "12345"
      },
      "features": {
        "my_feature_1": 1,
        "my_feature_2": 0.5
        ...
      }
    }
  ]
}