[Python] How to handle optional/null fields when inserting rows into TimeSeries container?

I'm using the griddb_python client to insert sensor data into a TimeSeries container in GridDB Cloud. The data comes from various IoT devices, and not all fields are always present in the payload.

Here’s an example of my payload:

{
"device_id": "sensor_A23",
"timestamp": "2025-06-12T10:00:00Z",
"temperature": 26.7,
"humidity": null
}



Container schema:

store.put_container("iot_data", [
("timestamp", griddb.Type.TIMESTAMP),
("device_id", griddb.Type.STRING),
("temperature", griddb.Type.FLOAT),
("humidity", griddb.Type.FLOAT)
], griddb.ContainerType.TIME_SERIES, True)



I’m trying to insert this row using:

container.put_row([
timestamp, device_id, temperature, humidity # humidity is None
])



But this raises an error:

griddb.gsexception.GSException: [ILLEGAL_PARAMETER(0x0123)] NULL is not allowed for column (Column index = 3)



Question:

Does GridDB support NULL/None values for numeric types in TimeSeries containers? If not, how should I handle missing sensor fields without using placeholder values like -999?

What I’ve tried:

• 
  Tried omitting the field from the row — raises an argument count mismatch error.
  
  
• 
  I checked the griddb_python GitHub repo and docs but no mention of how nullability works for numeric columns.
  
  
• 
  Replacing None with float('nan')—this is accepted, but it’s unclear if GridDB treats NaN as a valid missing value example for aggregates like AVG, SUM.
  

Continue reading...