You won’t be able to insert data into a Databricks table by sending an SQL INSERT statement in the HTTP body — that approach doesn’t work with Databricks REST APIs.
Databricks does not execute raw SQL from generic HTTP requests, even if Parse JSON succeeds. The API will accept the request but silently won’t write data, which is what you’re seeing.
Why your current approach fails
-
INSERT INTO table VALUES (…) in HTTP body is not supported
-
Parse JSON only validates the response structure
-
Databricks REST endpoints require specific APIs, not free-form SQL
-
SQL execution works only through:
So Power Apps → Custom Connector → “INSERT SQL” will never persist data.
Working methods (used in real projects)
✅ Method 1 — Databricks SQL Statement Execution API (recommended)
Use the Databricks SQL REST API:
POST /api/2.0/sql/statements
Example body:
{
"statement": "INSERT INTO my_schema.my_table VALUES ('101','John','90')",
"warehouse_id": "xxxx-warehouse-id"
}
Requirements:
-
SQL Warehouse must be running
-
Token must have SQL permissions
-
Table must be Delta table
This is the only supported way to run INSERT statements through REST.
✅ Method 2 — Create a Databricks Job (most stable)
-
Create a Databricks notebook:
data = {
"EmpId": dbutils.widgets.get("EmpId"),
"Name": dbutils.widgets.get("Name"),
"Score": dbutils.widgets.get("Score")
}
spark.createDataFrame([data]).write.mode("append").saveAsTable("my_table")
-
Call it from Power Automate / Custom Connector using:
POST /api/2.1/jobs/run-now
Pass values as notebook parameters.
This method is:
✅ Method 3 — Write to ADLS → Auto Loader (best architecture)
Flow:
Power Apps
↓
Power Automate
↓
ADLS Gen2 (JSON / CSV)
↓
Databricks Auto Loader
↓
Delta Table
This is how Databricks ingestion is normally designed.
Advantages:
-
no API throttling
-
high volume support
-
full retry handling
-
best performance
What is NOT supported
❌ Insert SQL via generic HTTP body
❌ Parse JSON → insert logic
❌ Direct JDBC from Power Apps
❌ Databricks REST without SQL warehouse
Recommended approach
For Power Apps integrations:
-
Small / real-time writes → SQL Statement API
-
Enterprise / bulk ingestion → ADLS + Auto Loader
-
Complex logic → Databricks job / notebook
Final answer
Your issue isn’t with Power Apps or the connector — Databricks simply does not support executing raw INSERT statements through standard HTTP APIs.
You must use either:
Once you switch to one of these, inserts will work correctly.
Share
Report
26
All responses (
Answers (
