Short answer: you can't switch this on from the wizard or a workspace setting. The Azure docs say the PostgreSQL connector is in Public Preview and that you request access through your Azure Databricks account team [1]. "This mode is not available in this workspace for the selected connector" means the preview isn't enabled for your workspace yet. Your support ticket is fine, but also ping your account team (or ask support to route it to them), and include the workspace ID and region.
While you wait, get the PostgreSQL side ready. Then the pipeline works on the first try once access lands.
1. Turn on logical replication
The connector reads changes through logical replication, PostgreSQL 13 or later [2].
SHOW wal_level; -- must return: logical
On Azure Database for PostgreSQL Flexible Server, set wal_level = logical in Server parameters (portal or CLI) and restart the server. Also open the firewall to your Databricks workspace [2].
2. Create the replication user, publication and slot
CREATE USER databricks_replication WITH PASSWORD '<secret>';
GRANT CONNECT ON DATABASE your_db TO databricks_replication;
GRANT USAGE ON SCHEMA your_schema TO databricks_replication;
GRANT SELECT ON TABLE your_schema.your_table TO databricks_replication;
ALTER USER databricks_replication WITH REPLICATION;
-- as the table owner, not the replication user
CREATE PUBLICATION databricks_publication FOR TABLE your_schema.your_table;
-- the slot must be created BY the replication user, pgoutput only
SET ROLE databricks_replication;
SELECT pg_create_logical_replication_slot('databricks_slot', 'pgoutput');
RESET ROLE;Order matters: create the publication first, then the slot [2]. You need one publication and one slot per database.
3. Don't skip the WAL limits
Set max_slot_wal_keep_size to a finite value; the default -1 lets an idle slot grow WAL without limit. Make sure max_replication_slots and max_wal_senders cover the new slot [2]. If you delete the pipeline later, drop the slot by hand.
4. Check before you retry the wizard
SHOW wal_level;
SELECT rolname, rolreplication FROM pg_roles WHERE rolname = 'databricks_replication';
SELECT * FROM pg_publication WHERE pubname = 'databricks_publication';
What I'd do
Ask the account team for PostgreSQL connector preview access on that workspace. In parallel, finish steps 1โ4. Once it is enabled, create the connection with the replication user's credentials and retry the wizard [3].
Note: I couldn't test this end-to-end because I don't have a PostgreSQL source wired to a workspace with the preview. Everything above comes from the current Azure docs.
References
[1] PostgreSQL ingestion connector | Azure Databricks โ https://learn.microsoft.com/en-us/azure/databricks/ingestion/lakeflow-connect/postgresql
[2] Configure PostgreSQL for ingestion into Azure Databricks โ https://learn.microsoft.com/en-us/azure/databricks/ingestion/lakeflow-connect/postgresql-source-setu...
[3] Create a PostgreSQL connection | Azure Databricks โ https://learn.microsoft.com/en-us/azure/databricks/ingestion/lakeflow-connect/postgresql-connection