The error you are encountering, "com.microsoft.sqlserver.jdbc.SQLServerException: Connection reset," even after a successful nc (netcat) connection, is a common but nuanced problem when connecting Databricks to an on-premise SQL Server. Although your notebook can ping the server and connect at the TCP/IP level, the JDBC handshake can still be interrupted or reset due to issues at the protocol, authentication, or firewall/proxy layer.​

Key Troubleshooting Steps

• Authentication Issues

  • Ensure that the username and password are valid for SQL Server authentication (not Windows Auth or Active Directory unless explicitly supported and configured).​

  • Confirm that SQL Authentication is enabled on the SQL Server.

• JDBC URL and Encryption

  • If encrypt=false was specified but SQL Server enforces SSL or vice versa, this can trigger a reset.

  • Try connecting with encrypt=true;trustServerCertificate=true in the JDBC string (unless your company policy or server prohibits this).​

  • Test with and without the encryption parameters, based on what the server expects.

• Firewall and Network Path

  • Even if ping/nc is successful, some firewalls allow TCP handshakes but block or reset application-layer traffic, especially over certain ports or for long-lived connections.​

  • Verify with your network team that the SQL Server port (default 1433) allows inbound JDBC traffic from your Databricks IPs for the full session duration.

  • Check for any "stateful inspection" or IDS/IPS devices that could drop or reset connections.

• SQL Server Configuration

  • Make sure TCP/IP is enabled on the SQL Server (SQL Server Configuration Manager > SQL Server Network Configuration > Protocols for [Instance]).​

  • Confirm that your SQL Server is bound to the correct IP or 'All IPs', and not 'localhost' only.

• JDBC Driver Version

  • Ensure you are using a compatible and up-to-date Microsoft JDBC driver for SQL Server. Old or mismatched drivers can cause handshake failures.

• Proxy/Gateway Issues

  • If connectivity to on-premise is via VPN, SSH tunnel, or another proxy, check that session timeouts/connection limits are not interfering.​

Additional Suggestions

• Use a simple test query (like SELECT 1) to rule out data or schema-related issues.

• Review SQL Server error logs for any incoming connection failures or network-level rejections.

• If possible, try the same JDBC connection from a different system on the same network to isolate if the problem is Databricks-specific.

• If your server enforces SSL and you provide encrypt=false, or vice versa, a reset is likely.​

• Some enterprise networks block dynamic port allocation; ensure the SQL Server is using a static port.​

Example JDBC URLs

For non-SSL (if server allows):

For SSL (try this if non-SSL does not work):

If custom instance (not the default), specify instanceName=....

What to Check Next

• Confirm the correct SQL authentication and connection parameters.

• Test with both encrypt=false and encrypt=true;trustServerCertificate=true.

• Review SQL Server and network/firewall logs for clues.

• Ask your network/security team if a proxy, NAT, or VPN is terminating the session unexpectedly.

Most likely, the issue is not basic TCP reachability but protocol-level, authentication, or enforced encryption mismatch. Work with your SQL Server and network teams to ensure all layers are compatible with your Databricks configurationration.​