Set up a Vertica source with epoch-based CDC

Keep the history for a day. Gluesync needs at least one hour (3600):

ALTER DATABASE DEFAULT SET HistoryRetentionTime = 86400;

Leave room for the sessions of the agent. The snapshot opens one connection for each parallel bucket of a table (four by default), and every entity needs sessions of its own for CDC:

ALTER DATABASE DEFAULT SET MaxClientSessions = 100;

Create the login that Gluesync uses. It is not a superuser and has no role. Replace the password:

CREATE USER gluesync IDENTIFIED BY '<a strong password>';
GRANT USAGE ON SCHEMA public TO gluesync;
GRANT USAGE ON SCHEMA sales TO gluesync;
GRANT SELECT ON SCHEMA sales TO gluesync;
-- Tables that are created later in the schema inherit the grants of the schema
ALTER SCHEMA sales DEFAULT INCLUDE PRIVILEGES;

Check the setting and the epochs:

SELECT parameter_name, current_value
FROM configuration_parameters
WHERE parameter_name IN ('HistoryRetentionTime', 'HistoryRetentionEpochs');

SELECT current_epoch, ahm_epoch, last_good_epoch FROM v_monitor.system;

In the Control Plane, select Create your first pipeline, or the + button of the pipeline list if you already have pipelines. The Agents setup wizard opens;

In 1. Setup source, search for vertica and choose the Vertica agent tile. Fill Base settings:

Select Test and continue. The source appears as a card with the Vertica logo and <user>@<host>:5433. Select Next;

In 2. Setup target(s), search for postgres and choose the PostgreSQL CDC agent tile. Fill the same five fields for your PostgreSQL database and select Test and continue, then Next;

In 3. Ready, enter a Pipeline name and select Finish.

In the pipeline, select Entities. The Add entities page opens with the Object browser. It lists the schemas that the login can use. Expand sales;

Select the table orders. The Field mapping dialog opens;

Under Target, choose the Target Schema public. In Target Table type orders and choose Create table "orders". The dialog shows New table orders will be created on confirm. and the proposed target types. The key column id is marked as a key: the dialog says that the table has keys inherited from the primary key of orders;

Select Confirm. The Create table orders dialog shows the Read-only SQL statement. Select Create table orders to run it on PostgreSQL;

Select the table customers and repeat steps 3 and 4;

Select Confirm and continue.

In the pipeline, find the entity sales.orders on Hold. Select the start options button, the one next to the play button. The Snapshot (copy-only) dialog opens;

Keep Keep and merge under Data already in the target. It overwrites rows with the same key and keeps the others. Keep the Standard speed, which uses one read and two writers;

Select Start snapshot. Do the same for sales.customers.

Select the play button of sales.orders. The Sync changes (CDC) dialog opens. Where it stopped is selected, because the snapshot left a checkpoint. Select Start entity. Do the same for sales.customers;

On Vertica, as the owner of the tables, insert, update and delete one row each:

INSERT INTO sales.orders (id, customer_id, amount, status, placed_at, note)
VALUES (1001, 2, 99.90, 'NEW', '2026-11-04 10:00:00+00', 'cdc insert');
COMMIT;

UPDATE sales.orders SET status = 'PAID', amount = 11.00 WHERE id = 1;
DELETE FROM sales.orders WHERE id = 2;
COMMIT;

After a few seconds, read the three keys on PostgreSQL:

SELECT id, status, amount, note FROM public.orders WHERE id IN (1, 2, 1001) ORDER BY id;
SELECT count(*) FROM public.orders;

Change a key. On Vertica run:

UPDATE sales.customers SET id = 30 WHERE id = 3;
COMMIT;

Truncate a table. TRUNCATE removes the history of the table together with its rows, so CDC cannot compare epochs any more. On Vertica run:

TRUNCATE TABLE sales.orders;
INSERT INTO sales.orders (id, customer_id, amount, status, placed_at, note)
VALUES (2001, 1, 1.00, 'NEW', '2026-11-04 13:00:00+00', 'after truncate'),
       (2002, 2, 2.00, 'NEW', '2026-11-04 13:00:00+00', 'after truncate');
COMMIT;

After a few seconds read SELECT id FROM public.orders ORDER BY id; on PostgreSQL and look for the warning in the Core Hub log:

docker logs <core hub container> 2>&1 | grep "cannot be read as changes"

In the pipeline list, select the Vertica icon of the pipeline. The Edit source page opens. Select Advanced settings;

Select Add property and open the Key list. It offers these settings:

Choose Snapshot read parallelism. The Value starts at 4. Enter 8 and select Test and continue.

Set HistoryRetentionTime to 60, add an entity and select Confirm and continue. The page shows The entity could not be saved. and the notification Entity Upsert Failed with The history of the database spans 60 seconds (HistoryRetentionTime), too little to read changes from: Gluesync needs at least 3600. A superuser sets it with ALTER DATABASE DEFAULT SET HistoryRetentionTime = 3600; Fix: run the statement of step 1;

Add a table with an array column, such as ARRAY[VARCHAR(20)], and let Gluesync create the PostgreSQL table. The Field mapping dialog proposes the type array, and Create table fails with Create target table execution error: ERROR: syntax error at or near "array". Fix: create the table yourself with the array type, tags text[], and choose it in Target Table. The mapping then shows array on both sides and the values arrive as {x,y};

Change the password in Edit source to a wrong value and select Test and continue. The page shows Failed to save: The connection was saved, but agent <id> of pipeline <id> did not accept the new credentials: Failed to initialize pool: [Vertica][VJDBC](2248) FATAL: Authentication failed for username "gluesync". The wrong password is stored. Fix: enter the right one and select Test and continue again. The pipeline recovers by itself.