Meilisearch¶

Index events as searchable documents in Meilisearch.

Docker Image¶

docker pull ghcr.io/psteinroe/postgres-stream:meilisearch-latest

Configuration¶

sink:
  type: meilisearch
  url: http://localhost:7700
  index: events

With Authentication¶

sink:
  type: meilisearch
  url: http://localhost:7700
  index: events
  api_key: ${MEILISEARCH_API_KEY}

Options¶

Dynamic Routing¶

Route events to different indexes using metadata:

-- Route by table name
metadata_extensions = '[
  {"json_path": "index", "expression": "tg_table_name"}
]'

-- Or use static metadata
metadata = '{"index": "products"}'

The sink reads index from event metadata.

Primary Key¶

Meilisearch requires each document to have a primary key. Options:

1. Configure in Meilisearch: Set the primary key field when creating the index
2. Add via payload_extensions: Transform the event ID into the expected field

-- Add id field from event
payload_extensions = '[
  {"json_path": "id", "expression": "new.id::text"}
]'

Document Format¶

The event payload is indexed directly (not the full envelope). Your payload structure should match your Meilisearch schema.

Task Handling¶

The sink waits for indexing tasks to complete before returning success. This ensures data consistency but may impact throughput for large batches.

Example¶

Complete configuration:

stream:
  id: 1
  pg_connection:
    host: localhost
    port: 5432
    name: mydb
    username: postgres
    password: postgres
    tls:
      enabled: false
  batch:
    max_size: 1000
    max_fill_secs: 5

sink:
  type: meilisearch
  url: http://localhost:7700
  index: products
  api_key: masterKey

Index Setup¶

Create an index with appropriate settings:

curl -X POST 'http://localhost:7700/indexes' \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer masterKey' \
  --data-binary '{
    "uid": "products",
    "primaryKey": "id"
  }'