Skip to main content
This guide shows you how to capture Postgres changes with Sequin and stream them to an Elasticsearch index. With Postgres data flowing into Elasticsearch you can power full‑text search, vector retrieval, and analytics without manual re‑indexing. By the end of this guide you’ll have an Elasticsearch index that stays continuously in sync with your database.
This is the how-to guide for streaming Postgres to Elasticsearch. See the quickstart for a step-by-step walkthrough or the reference for details on all configuration options.

​
Prerequisites

If you self‑host Sequin:
  1. Install Sequin
  2. Connect a Postgres database
If you use Sequin Cloud:
  1. Sequin installed
  2. Connect a Postgres database

​
Basic setup

​
Prepare an Elasticsearch cluster

Sequin converts Postgres rows to JSON documents and sends them to the Elasticsearch Bulk API. You need a reachable cluster:
  • Local development ⇒ run Elasticsearch in Docker: 
    curl -fsSL https://elastic.co/start-local | sh
    The script starts Elasticsearch + Kibana and prints connection details: 
    🔌 Elasticsearch API endpoint: http://localhost:9200
🔑 API key: <api-key>
    Use the API key (not the user/password) when configuring the sink.
  • Production ⇒ Elastic Cloud or self‑managed cluster.
If your Elasticsearch instance is running on your laptop but Sequin runs in the cloud, connect them with the Sequin CLI’s tunnel command.

​
Map Postgres tables to Elasticsearch indices

Elasticsearch stores documents in indices. As a rule of thumb create one sink per Postgres table → one index. That keeps the mapping consistent and simplifies queries. Advanced scenarios – sharding, multi‑tenancy, or routing documents to multiple indices – can be handled with filters and soon with Sequin Routing functions.

​
Create an Elasticsearch sink

Navigate to Sinks → Create sink → Elasticsearch and follow the steps below.
1

Source configuration

Under Source pick the table or schema you want to index in Elasticsearch.Optionally:
  • Select which actions (insert, update, delete) you care about.
  • Add filters such as in_stock = true to index only a subset of rows.
2

Backfill existing rows

Enable Initial backfill if you want Sequin to load existing rows into Elasticsearch before streaming live changes.
3

Transform rows to documents

Your transform must emit JSON compatible with the index mapping. Sequin automatically builds the document _id from the table’s primary key. Let us know if you need to use a different _id field.Typical transform:
def transform(_action, record, _changes, _metadata) do
  %{
    name: record["name"],
    description: record["description"],
    price: record["price"],
    in_stock: record["in_stock"]
  }
end
For dynamic schemas you can just return record unchanged. See the Elasticsearch ingest best‑practices for mapping guidance.
4

Delivery settings

In the Elasticsearch card enter:
  • Endpoint URL: http://host.docker.internal:9200 (or your cluster URL)
  • Index name: products
  • Authentication type: api_key
  • Authentication value: <api-key>
Keep the default Batch size unless you have special throughput needs. Sequin supports up to 10,000 docs per batch.
5

Create the sink

Name the sink (e.g. products-elasticsearch) and click Create sink. Sequin queues a backfill (if selected) and then starts streaming live changes.

​
Verify & debug

  1. In Sequin’s web console watch the Messages count increase.
  2. Query Elasticsearch: 
    curl -X GET "http://localhost:9200/products/_search?pretty" \
  -H "Authorization: ApiKey <api-key>" \
  -H "Content-Type: application/json" \
  -d '{"query": {"match_all": {}}}'
  3. If documents are missing:
    • Check the Messages tab for failed deliveries.
    • Inspect the error returned by Elasticsearch; mapping conflicts and authentication issues are common.

​
Next steps

Ready to go further? Explore: