Create VectorDB Connection

Endpoint

POST /apiops/projects/{projectName}/vector-dbs/{vectorDbName}/

Authentication

Requires a Personal API Access Token.

Header

Authorization: Bearer YOUR_TOKEN

Request

Headers

Header	Value	Required
Authorization	Bearer {token}	Yes
Content-Type	application/json	Yes

Path Parameters

Parameter	Type	Required	Description
projectName	string	Yes	Project name
vectorDbName	string	Yes	Vector DB connection name (must equal name in the body)

Query Parameters

None

Request Body

The body is a ConnectionConfig of subtype vector-db, so it must include "_class": "vector-db" at the top level. The connection details live in the nested transport object, whose own "family" discriminator (jdbc / http / redis) is selected by dbType.

Full JSON Body Example - JDBC transport (pgvector)

{
  "_class": "vector-db",
  "name": "pgvector-primary",
  "description": "Primary pgvector store for RAG",
  "enabled": true,
  "dbType": "PGVECTOR",
  "defaultCollectionName": "documents",
  "embeddingDimension": 1536,
  "distanceMetric": "COSINE",
  "maxBatchSize": 100,
  "transport": {
    "family": "jdbc",
    "connectTimeoutMs": 5000,
    "requestTimeoutMs": 30000,
    "jdbcUrl": "jdbc:postgresql://db.internal:5432/vectors",
    "username": "apinizer",
    "password": "s3cr3t-db-pass",
    "pool": {
      "minimumIdle": 2,
      "maximumPoolSize": 5,
      "connectionTimeoutMs": 30000,
      "idleTimeoutMs": 600000,
      "maxLifetimeMs": 1800000
    }
  }
}

Full JSON Body Example - HTTP transport (Qdrant / Pinecone / Weaviate / Milvus / Chroma)

{
  "_class": "vector-db",
  "name": "qdrant-cloud",
  "description": "Qdrant Cloud cluster",
  "enabled": true,
  "dbType": "QDRANT",
  "defaultCollectionName": "documents",
  "embeddingDimension": 1536,
  "distanceMetric": "COSINE",
  "maxBatchSize": 100,
  "transport": {
    "family": "http",
    "connectTimeoutMs": 5000,
    "requestTimeoutMs": 30000,
    "endpoint": "https://xyz.eu-central.aws.cloud.qdrant.io:6333",
    "apiVersion": "v1",
    "authScheme": "API_KEY_HEADER",
    "authHeaderName": "api-key",
    "apiKey": "qdrant-api-key-value",
    "regionCode": "eu-central"
  }
}

Full JSON Body Example - Redis transport (RediSearch)

{
  "_class": "vector-db",
  "name": "redis-search",
  "description": "RediSearch vector index",
  "enabled": true,
  "dbType": "REDIS",
  "defaultCollectionName": "doc_idx",
  "embeddingDimension": 1536,
  "distanceMetric": "COSINE",
  "maxBatchSize": 100,
  "transport": {
    "family": "redis",
    "connectTimeoutMs": 5000,
    "requestTimeoutMs": 30000,
    "uri": "rediss://redis.internal:6379",
    "username": "default",
    "password": "redis-auth-pass"
  }
}

Request Body Fields

Field	Type	Required	Default	Description
_class	string	Yes	-	Connection subtype discriminator — must be "vector-db"
name	string	Yes	-	Connection name — must match vectorDbName in the path
description	string	No	-	Free-text description
enabled	boolean	No	true	Whether the connection is enabled
dbType	enum	Yes	-	Vector store type: PGVECTOR, QDRANT, WEAVIATE, MILVUS, CHROMA, PINECONE, REDIS, OTHER
defaultCollectionName	string	No	-	Default collection / index / table name
embeddingDimension	integer	No	-	Embedding vector dimension (e.g. 1536)
distanceMetric	enum	No	vendor default	Distance metric: COSINE, EUCLIDEAN, DOT_PRODUCT. Filled from dbType when omitted
maxBatchSize	integer	No	vendor default	Maximum upsert/query batch size. Filled from dbType when omitted
transport	object	Yes	-	Polymorphic transport sub-document (see below)

transport (family jdbc) — pgvector

Field	Type	Required	Description
family	string	Yes	Transport discriminator — "jdbc"
connectTimeoutMs	integer	No	Connect timeout (ms). null → vendor client default
requestTimeoutMs	integer	No	Request timeout (ms). null → vendor client default
jdbcUrl	string	Yes	JDBC URL, e.g. jdbc:postgresql://host:port/db
username	string	No	Database username
password	string	No	Database password (secret — masked on read; blank on update preserves the stored value)
pool	object	No	HikariCP pool settings (all fields nullable → HikariCP defaults)
pool.minimumIdle	integer	No	Minimum idle connections
pool.maximumPoolSize	integer	No	Maximum pool size
pool.connectionTimeoutMs	integer	No	Connection acquisition timeout (ms)
pool.idleTimeoutMs	integer	No	Idle connection timeout (ms)
pool.maxLifetimeMs	integer	No	Maximum connection lifetime (ms)

transport (family http) — Qdrant / Pinecone / Weaviate / Milvus / Chroma

Field	Type	Required	Description
family	string	Yes	Transport discriminator — "http"
connectTimeoutMs	integer	No	Connect timeout (ms)
requestTimeoutMs	integer	No	Request timeout (ms)
endpoint	string	Yes	REST endpoint URL
apiVersion	string	No	Vendor API version
authScheme	enum	No	Auth scheme: BEARER, API_KEY_HEADER, BASIC, NONE, CUSTOM
authHeaderName	string	No	Header name for API-key auth (used with API_KEY_HEADER)
apiKey	string	No	API key (secret — masked on read; blank on update preserves the stored value)
apiSecret	string	No	API secret (secret — masked on read; blank on update preserves the stored value)
regionCode	string	No	Vendor region code (e.g. Pinecone)

transport (family redis) — RediSearch

Field	Type	Required	Description
family	string	Yes	Transport discriminator — "redis"
connectTimeoutMs	integer	No	Connect timeout (ms)
requestTimeoutMs	integer	No	Command timeout (ms)
uri	string	Yes	Redis URI (redis:// / rediss://; bare host:port also accepted)
username	string	No	Redis ACL username (Redis 6+)
password	string	No	Redis AUTH password (secret — masked on read; blank on update preserves the stored value)

Notes

• vectorDbName in the path must equal name in the body (case-insensitive), otherwise 400 Bad Request is returned.
• Sending a name that already exists in the project updates that connection in place (upsert); a new name creates a new connection.

Response

Success Response (200 OK)

{
  "status": "SUCCESS",
  "deploymentResult": {
    "success": true
  }
}

Error Response (400 Bad Request)

{
  "status": "FAILURE",
  "resultMessage": "Vector DB connection name in path (pgvector-primary) does not match name in body (pgvector-2)!"
}

or

{
  "status": "FAILURE",
  "resultMessage": "Vector DB connection body can not be empty!"
}

Error Response (401 Unauthorized)

{
  "status": "FAILURE",
  "resultMessage": "Token is not valid!"
}

cURL Example

curl -X POST \
  "https://demo.apinizer.com/apiops/projects/MyProject/vector-dbs/pgvector-primary/" \
  -H "Authorization: Bearer YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "_class": "vector-db",
    "name": "pgvector-primary",
    "description": "Primary pgvector store for RAG",
    "enabled": true,
    "dbType": "PGVECTOR",
    "defaultCollectionName": "documents",
    "embeddingDimension": 1536,
    "distanceMetric": "COSINE",
    "maxBatchSize": 100,
    "transport": {
      "family": "jdbc",
      "jdbcUrl": "jdbc:postgresql://db.internal:5432/vectors",
      "username": "apinizer",
      "password": "s3cr3t-db-pass"
    }
  }'

Notes and Warnings

• Upsert: POST /{vectorDbName}/ creates the connection, or updates it in place if the name already exists in the project.
• Discriminators: The body must carry "_class": "vector-db" at the top level, and the nested transport must carry its own "family" discriminator matching the dbType family (jdbc for pgvector, http for Qdrant/Pinecone/Weaviate/Milvus/Chroma, redis for RediSearch).
• Secrets: Secret fields (JDBC password, HTTP apiKey/apiSecret, Redis password) are encrypted server-side on save and returned masked on read.
• Save = encrypt + deploy: A successful save encrypts @SecretData fields and pushes the connection to the workers.
• Data-plane Defaults: distanceMetric and maxBatchSize are auto-filled from dbType when omitted (existing user values are preserved).

Related Documentation

• Update VectorDB Connection - Update an existing connection (blank-secret preserve)
• Test VectorDB Connection - Validate connectivity before saving
• VectorDB Connections API - Resource overview
• AI Gateway API - Full AI Gateway APIops surface