Facing difficulties in accessing Feast online store

[VidhyaPandi]

We are trying to use Feast as the feature store for our analytics platform. Offline store and online store were successfully setup and we are able to access via Jupyter notebook using Feast python SDK. However, we are facing difficulties in accessing Feast online store.
Issue Details:

1. We tried setting up a feature server in Kubernetes environment to serve online features via REST API. We can invoke the endpoint /get-online-features with status 200 OK. But the response contains only NULL values as shown below.
   {
   "metadata": {
   "feature_names": [
   "equipment_id",
   "usage_metrics"
   ]
   },
   "results": [
   {
   "values": [
   14991,
   14989
   ],
   "statuses": [
   "PRESENT",
   "PRESENT"
   ],
   "event_timestamps": [
   "1970-01-01T00:00:00Z",
   "1970-01-01T00:00:00Z"
   ]
   },
   {
   "values": [
   null,
   null
   ],
   "statuses": [
   "NOT_FOUND",
   "NOT_FOUND"
   ],
   "event_timestamps": [
   "1970-01-01T00:00:00Z",
   "1970-01-01T00:00:00Z"
   ]
   }
   ]
   }

2. To run an inference in SPARK environment we tried to query the postgres database (online store) to fetch the features to feed the inference.
   We still get the null values in value column.
   Query:
   SELECT entity_key, feature_name,value FROM feast_features
   LIMIT 5;
   Deserialization function: (udf)
   def deserialize_value(serialized_value):
   if serialized_value is None:
   return None # Handle null values safely

   if isinstance(serialized_value, bytearray):
   serialized_value = bytes(serialized_value) # Convert bytearray to bytes

   Deserialize using Protobuf

   value_proto = Value_pb2.Value()
   value_proto.ParseFromString(serialized_value)

   Extract the actual feature value

   if value_proto.HasField("int64_val"):
   return value_proto.int64_val
   elif value_proto.HasField("double_val"):
   return value_proto.double_val
   elif value_proto.HasField("string_val"):
   return value_proto.string_val
   elif value_proto.HasField("bool_val"):
   return value_proto.bool_val
   elif value_proto.HasField("bytes_val"):
   return value_proto.bytes_val.decode("utf-8") # Convert bytes to string
   else:
   return None

We decided to go with postgres as datasource so we installed Feast using
pip install feast['postgres']
Now the feature store set up has been completed, We are able to get the online features using feast SDK and here is configuration details.
feature_store.yaml:

project: project_name
provider: local
registry:
registry_type: sql
path: postgresql+psycopg2://user:[email protected]:55001/feast_db
cache_ttl_seconds: 60
sqlalchemy_config_kwargs:
echo: false
pool_pre_ping: true
online_store:
type: postgres
host: host_ip1
port: port1
database: db_name
db_schema: schema_name
user: user1
password: password1
offline_store:
type: postgres
host: host_ip2
port: port2
database: db_name
db_schema: schema_name
user: user2
password: password2
entity_key_serialization_version: 2
auth:
type: no_auth

Please help us to integrate Feast online store with our ML ops inference pipeline running in SPARK cluster. So that we could pass the latest data from online store for predicting ML models via pipeline.

[redhatHameed]

can you verify if you have performed materialization successfully and have data in onlinestore db ?

[VidhyaPandi]

Yes, I did materialization and got the results via SDK. Please refer to the screenshot below: [cid:49b2e3fa-1245-41bc-943d-cef4c5cea949]

…

________________________________ From: Abdul Hameed ***@***.***> Sent: Thursday, January 30, 2025 11:53 PM To: feast-dev/feast ***@***.***> Cc: Vidhya Pandi ***@***.***>; Author ***@***.***> Subject: Re: [feast-dev/feast] Facing difficulties in accessing Feast online store (Issue #4986) can you verify if you have performed materialization successfully and have data in onlinestore db ? — Reply to this email directly, view it on GitHub<#4986 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/BIBPKIC4GSG6URILSFZ6IPL2NJU3VAVCNFSM6AAAAABWE5XY6GVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMMRVGI2TIMRYHE>. You are receiving this because you authored the thread.Message ID: ***@***.***> CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe. Disclaimer: This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you are not the intended recipient you are notified that disclosing, copying, distributing or taking any action in reliance on the contents of this information is strictly prohibited.