❗ Further development is discontinued. In collaboration with the Feast community this repo is now added to the main project. From v0.19 feast-dev/feast#2349 |
---|
This repo contains a plugin for feast to run an offline store on Spark.
It can be installed from pip and configured in the feature_store.yaml
configuration file to interface with DataSources
using Spark.
Note that this repository has not yet had a major release as it is still work in progress.
We strongly encourage you to contribute to our repository. Find out more in our contribution guidelines
Currently only supports Feast versions >=0.15.0
.
pip install feast-spark-offline-store
Or to install from source:
git clone git@github.com:Adyen/feast-spark-offline-store.git
cd feast-spark-offline-store
pip install -e '.[dev]'
Install feast
and feast_spark_offline_store
and change the Feast configurations in feature_store.yaml
to use feast_spark_offline_store.SparkOfflineStore
:
project: example_feature_repo
registry: data/registry.db
provider: local
online_store:
...
offline_store:
type: feast_spark_offline_store.spark.SparkOfflineStore
spark_conf:
spark.master: "local[*]"
spark.ui.enabled: "false"
spark.eventLog.enabled: "false"
spark.sql.catalogImplementation: "hive"
spark.sql.parser.quotedRegexColumnNames: "true"
spark.sql.session.timeZone: "UTC"
See Feast documentation on offline stores and adding custom offline stores.
MIT license. For more information, see the LICENSE file.