pg-trickle Integration

pg-trickle is the PostgreSQL extension that pg_tide was originally extracted from. If you already use pg-trickle, you can adopt pg_tide incrementally — or run both side-by-side during a migration.

When to use pg_tide instead of pg-trickle

Schema compatibility

pg_tide uses a tide.* schema prefix. pg-trickle uses pg_trickle_* table names. The two schemas can coexist in the same database without conflict.

Migrating from pg-trickle

1. Install pg_tide alongside pg-trickle

CREATE EXTENSION pg_tide;

2. Create matching outboxes and inboxes

For each pg_trickle_outbox in your existing schema:

SELECT tide.outbox_create(outbox_name, retention_hours)
FROM pg_trickle_outbox_config;

3. Migrate pending messages

INSERT INTO tide.tide_outbox_messages (outbox_name, payload, headers, created_at)
SELECT stream_name, payload, headers, created_at
FROM pg_trickle_outbox_messages
WHERE consumed_at IS NULL;

4. Point the relay at pg_tide

Update your pg-tide relay configuration:

SELECT tide.relay_set_outbox(
    'my-pipeline',
    'orders',
    'nats',
    '{"url":"nats://broker:4222"}'::jsonb
);

5. Verify and cut over

Run both relays in parallel during the transition. Once the pg_tide relay is processing all new messages, decommission the pg-trickle relay.

Using both together

Both extensions can write to the same NATS subject or Kafka topic — consumers should use the x-source header to distinguish messages originating from pg_tide versus pg-trickle.

pg_tide stamps every envelope with:

{
  "id": 42,
  "outbox_name": "orders",
  "payload": { ... },
  "headers": { "x-source": "pg-tide", "x-outbox": "orders" }
}