Connector API tutorial

Learn how to set up a self-managed connector client using the Elasticsearch Connector APIs.

For this example we’ll use the PostgreSQL connector to sync data from a PostgreSQL database to Elasticsearch. We’ll spin up a simple PostgreSQL instance in Docker with some example data, create a connector, and sync the data to Elasticsearch. You can follow the same steps to set up a connector for another data source.

If you already have an Elasticsearch deployment on Elastic Cloud (Hosted deployment or Serverless project), you’re good to go. To spin up Elasticsearch in local dev mode in Docker for testing purposes, open the collapsible section below.

docker run -p 9200:9200 -d --name elasticsearch \
  -e "discovery.type=single-node" \
  -e "xpack.security.enabled=false" \
  -e "xpack.security.http.ssl.enabled=false" \
  -e "xpack.license.self_generated.type=trial" \
  docker.elastic.co/elasticsearch/elasticsearch:8.15.3

export ELASTIC_PASSWORD="changeme"

curl -s -X GET -u elastic:$ELASTIC_PASSWORD http://localhost:9200

curl -s -X PUT http://localhost:9200/_connector/my-connector-id \
-H "Authorization: APIKey $APIKEY" \
-H "Content-Type: application/json" \
-d '{
  "name": "Music catalog",
  "index_name":  "music",
  "service_type": "postgresql"
}'

For this tutorial, we’ll set up a PostgreSQL instance in Docker with some example data. Of course, you can skip this step and use your own existing PostgreSQL instance if you have one. Keep in mind that using a different instance might require adjustments to the connector configuration described in the next steps.

docker run --name postgres -e POSTGRES_USER=myuser -e POSTGRES_PASSWORD=mypassword -p 5432:5432 -d postgres

mkdir -p ~/data

curl -L https://raw.githubusercontent.com/lerocha/chinook-database/master/ChinookDatabase/DataSources/Chinook_PostgreSql.sql -o ~/data/Chinook_PostgreSql.sql

docker cp ~/data/Chinook_PostgreSql.sql postgres:/
docker exec -it postgres psql -U myuser -f /Chinook_PostgreSql.sql

docker exec -it postgres psql -U myuser -d chinook -c "\dt"

This tutorial uses a very basic setup. To use advanced functionality such as filtering rules and incremental syncs, enable track_commit_timestamp on your PostgreSQL database. Refer to Tutorial for more details.

Now it’s time for the real fun! We’ll set up a connector to create a searchable mirror of our PostgreSQL data in Elasticsearch.

We’ll use the Create connector API to create a PostgreSQL connector instance.

Run the following API call, using the Dev Tools Console or curl:

PUT _connector/my-connector-id
{
  "name": "Music catalog",
  "index_name":  "music",
  "service_type": "postgresql"
}

Note that we specified the my-connector-id ID as a part of the PUT request. We’ll need the connector ID to set up and run the connector service locally.

If you’d prefer to use an autogenerated ID, replace PUT _connector/my-connector-id with POST _connector.

Now we’ll run the connector service so we can start syncing data from our PostgreSQL instance to Elasticsearch. We’ll use the steps outlined in Running from a Docker container.

When running the connectors service on your own infrastructure, you need to provide a configuration file with the following details:

If you haven’t already created an API key to access Elasticsearch, you can use the _security/api_key endpoint.

Here, we assume your target Elasticsearch index name is music. If you use a different index name, adjust the request body accordingly.

POST /_security/api_key
{
  "name": "music-connector",
  "role_descriptors": {
    "music-connector-role": {
      "cluster": [
        "monitor",
        "manage_connector"
      ],
      "indices": [
        {
          "names": [
            "music",
            ".search-acl-filter-music",
            ".elastic-connectors*"
          ],
          "privileges": [
            "all"
          ],
          "allow_restricted_indices": false
        }
      ]
    }
  }
}

You’ll need to use the encoded value from the response as the elasticsearch.api_key in your configuration file.

Let’s create a directory and a config.yml file to store the connector configuration:

mkdir -p ~/connectors-config
touch ~/connectors-config/config.yml

Now, let’s add our connector details to the config file. Open config.yml and paste the following configuration, replacing placeholders with your own values:

elasticsearch.host: <ELASTICSEARCH_ENDPOINT> # Your Elasticsearch endpoint
elasticsearch.api_key: <ELASTICSEARCH_API_KEY> # Your Elasticsearch API key

connectors:
  - connector_id: "my-connector-id"
    service_type: "postgresql"

We provide an example configuration file in the elastic/connectors repository for reference.

Now that we have the configuration file set up, we can run the connector service locally. This will point your connector instance at your Elasticsearch deployment.

Run the following Docker command to start the connector service:

docker run \
-v "$HOME/connectors-config:/config" \
--rm \
--tty -i \
--network host \
docker.elastic.co/enterprise-search/elastic-connectors:8.15.3.0 \
/app/bin/elastic-ingest \
-c /config/config.yml

Verify your connector is connected by getting the connector status (should be needs_configuration) and last_seen field (note that time is reported in UTC). The last_seen field indicates that the connector successfully connected to Elasticsearch.

GET _connector/my-connector-id

Now our connector instance is up and running, but it doesn’t yet know where to sync data from. The final piece of the puzzle is to configure our connector with details about our PostgreSQL instance. When setting up a connector in the Elastic Cloud or Serverless UIs, you’re prompted to add these details in the user interface.

But because this tutorial is all about working with connectors programmatically, we’ll use the Update connector configuration API to add our configuration details.

Run the following API call to configure the connector with our PostgreSQL configuration details:

PUT _connector/my-connector-id/_configuration
{
  "values": {
    "host": "127.0.0.1",
    "port": 5432,
    "username": "myuser",
    "password": "mypassword",
    "database": "chinook",
    "schema": "public",
    "tables": "album,artist"
  }
}

We’re now ready to sync our PostgreSQL data to Elasticsearch. Run the following API call to start a full sync job:

POST _connector/_sync_job
{
    "id": "my-connector-id",
    "job_type": "full"
}

To store data in Elasticsearch, the connector needs to create an index. When we created the connector, we specified the music index. The connector will create and configure this Elasticsearch index before launching the sync job.

Use the Get sync job API to track the status and progress of the sync job. By default, the most recent job statuses are returned first. Run the following API call to check the status of the sync job:

GET _connector/_sync_job?connector_id=my-connector-id&size=1

The job document will be updated as the sync progresses, you can check it as often as you’d like to poll for updates.

Once the job completes, the status should be completed and indexed_document_count should be 622.

Verify that data is present in the music index with the following API call:

GET music/_count

Elasticsearch stores data in documents, which are JSON objects. List the individual documents with the following API call:

GET music/_search

Use the following command to inspect the latest sync job’s status:

GET _connector/_sync_job?connector_id=my-connector-id&size=1

If the connector encountered any errors during the sync, you’ll find these in the error field.

To delete the connector and its associated sync jobs run this command:

DELETE _connector/my-connector-id&delete_sync_jobs=true

This won’t delete the Elasticsearch index that was created by the connector to store the data. Delete the music index by running the following command:

DELETE music

To remove the PostgreSQL container, run the following commands:

docker stop postgres
docker rm postgres

To remove the connector service, run the following commands:

docker stop <container_id>
docker rm <container_id>

Congratulations! You’ve successfully set up a self-managed connector using the Connector APIs.

Here are some next steps to explore: