Actions

The ballerinax/milvus package exposes the following clients:

Client	Purpose
Client	Manage collections, indexes, and vector data; perform similarity search and query operations.

---

Client

Manage collections, indexes, and vector data; perform similarity search and query operations.

Configuration

Field	Type	Default	Description
authConfig	AuthConfig	()	Token-based authentication configuration.
credentialsConfig	CredentialsConfig	()	Username and password authentication configuration.
idleTimeout	int	()	Idle timeout in seconds.
keepAliveTime	int	55	Keep-alive ping interval in seconds.
keepAliveTimeout	int	20	Keep-alive timeout in seconds.
keepAliveWithoutCalls	boolean	false	Whether to send keep-alive pings without active calls.
rpcDeadline	int	0	RPC deadline in seconds. 0 means no deadline.
connectTimeout	int	10	Connection timeout in seconds.
databaseName	string	()	Target database name.
serverName	string	()	Server name for TLS verification.
proxyAddress	string	()	Proxy server address.
secureConfig	SecureConfig	()	TLS/SSL configuration with client and server certificate paths.

Initializing the client

import ballerinax/milvus;

configurable string serviceUrl = ?;
configurable string token = ?;

milvus:Client milvusClient = check new (serviceUrl, {
    authConfig: {
        token: token
    }
});

Operations

Collection management

listCollections

Lists the names of all collections in the connected Milvus instance.

Returns: string[]|Error

Sample code:

string[] collections = check milvusClient->listCollections();

Sample response:

["products", "documents", "image_embeddings"]

createCollection

Creates a new collection with the specified name, primary field, and vector dimension.

Parameters:

Name	Type	Required	Description
request	CreateCollectionRequest	Yes	Collection creation parameters including name, primary field name, and vector dimension.

Returns: Error?

Sample code:

check milvusClient->createCollection({
    collectionName: "products",
    primaryFieldName: "id",
    dimension: 768
});

loadCollection

Loads a collection into memory to enable search and query operations.

Parameters:

Name	Type	Required	Description
collectionName	string	Yes	The name of the collection to load.

Returns: Error?

Sample code:

check milvusClient->loadCollection("products");

Index management

createIndex

Creates indexes on the specified fields of a collection for efficient search.

Parameters:

Name	Type	Required	Description
request	CreateIndexRequest	Yes	Index creation parameters including collection name, primary key, and field names to index.

Returns: Error?

Sample code:

check milvusClient->createIndex({
    collectionName: "products",
    primaryKey: "id",
    fieldNames: ["vector"]
});

Data operations

upsert

Inserts a new vector entry or updates it if the primary key already exists.

Parameters:

Name	Type	Required	Description
request	UpsertRequest	Yes	Upsert parameters including collection name, vector data, optional primary key, and properties.

Returns: Error?

Sample code:

check milvusClient->upsert({
    collectionName: "products",
    data: {
        primaryKey: {
            fieldName: "id",
            value: 1
        },
        vectors: [0.12, 0.34, 0.56, 0.78],
        properties: {
            "name": "Wireless Headphones",
            "category": "Electronics"
        }
    }
});

delete

Deletes vectors from a collection by IDs or a filter expression. Returns the number of deleted entries.

Parameters:

Name	Type	Required	Description
request	DeleteRequest	Yes	Delete parameters including collection name and either IDs or a filter expression.

Returns: int|Error

Sample code:

int deletedCount = check milvusClient->delete({
    collectionName: "products",
    ids: [1, 2, 3]
});

Sample response:

3

Search & query

search

Performs a similarity search on vector data and returns the top-K most similar results.

Parameters:

Name	Type	Required	Description
request	SearchRequest	Yes	Search parameters including collection name, query vectors, top-K count, optional filter, and output fields.

Returns: SearchResult[][]|Error

Sample code:

milvus:SearchResult[][] results = check milvusClient->search({
    collectionName: "products",
    vectors: [0.12, 0.34, 0.56, 0.78],
    topK: 3,
    outputFields: ["name", "category"]
});

Sample response:

[[{"primaryKey": "1", "id": 1, "similarityScore": 0.98, "outputFields": null}, {"primaryKey": "5", "id": 5, "similarityScore": 0.87, "outputFields": null}, {"primaryKey": "12", "id": 12, "similarityScore": 0.76, "outputFields": null}]]

query

Queries vectors from a collection using a filter expression and returns matching entries.

Parameters:

Name	Type	Required	Description
request	QueryRequest	Yes	Query parameters including collection name, filter expression, and output fields.

Returns: QueryResult[][]|Error

Sample code:

milvus:QueryResult[][] results = check milvusClient->query({
    collectionName: "products",
    filter: "category == 'Electronics'",
    outputFields: ["name", "category"]
});

Sample response:

[[{"name": "Wireless Headphones", "category": "Electronics"}, {"name": "Bluetooth Speaker", "category": "Electronics"}]]