Databricks Test Connection Fails in Sparkflows – How to Fix?

Integration Databricks
1

Problem

While clicking Test Connection for Databricks in Sparkflows, I’m getting one of the following errors:

  • Not able to access the resource. Please check host url
  • Invalid access token. Please check token
  • HTTP failure response for http://<host_ip>:8080/testConnection: 403 Forbidden

How can I troubleshoot and fix this issue?

Solution

This issue usually occurs when Sparkflows is unable to communicate with Databricks or when the authentication details are incorrect. Below are the common causes and step-by-step troubleshooting actions.

Possible Causes

  1. The JDBC Databricks URL configured in Sparkflows is incorrect.
  2. The Databricks access token is invalid or misconfigured.
  3. The Databricks REST API is not accessible from the Sparkflows machine.
  4. The Databricks token does not have sufficient permissions.
  5. A Load Balancer / WAF (X-XSRF restriction) is blocking the request.

Troubleshooting Steps

Sparkflows connects to Databricks using the Databricks REST API to submit jobs. Ensure that this API is accessible from the Sparkflows host.

Step 1: Check Network Connectivity

Verify that the Sparkflows machine can reach the Databricks workspace:

telnet xxxx.cloud.databricks.com 443

Note: Replace xxxx.cloud.databricks.com with your actual Databricks Workspace URL.

Step 2: Validate Databricks Token & REST API Access

Test whether the Databricks endpoint is reachable and the token is valid by fetching cluster details:

curl -i https://xxxxx.azuredatabricks.net/api/2.0/clusters/get?cluster_id=xxxxx \
-H "Authorization: Bearer <Databricks_Token>"

Note:

  • Replace the Databricks host URL with your workspace URL
  • Update cluster_id with a valid cluster ID
  • Ensure the token has required permissions

Step 3: Test Databricks Connection from Sparkflows API

Use the Sparkflows API to test the Databricks connection:

curl --location 'http://sparkflows_host/api/v1/connections/test' \
--header 'token: sparkflows_token' \
--header 'Content-Type: application/json' \
--data '{
  "credentialStoreId": -1,
  "id": 1,
  "url": "JDBC Databricks URL",
  "password": "Databricks_Token",
  "connectionType": "databricks",
  "authenticationType": "USER_CREDENTIAL"
}'

Note: Replace the following with actual values:

  • sparkflows_host
  • sparkflows_token
  • JDBC Databricks URL
  • Databricks_Token

Step 4: Check WAF / Load Balancer Settings (Azure)

If you are using Azure Application Gateway with WAF v2:

  • Disable Inspect Request Body in Global Parameters
  • This avoids request blocking due to X-XSRF restrictions