MilvusClientV2

A ClientV2 instance represents a Rust client that connects to a specific Milvus instance and exposes the request/response-based Milvus API.

Import the client and its entry points with the prelude:

use milvus::v2::prelude::*;

The prelude re-exports ClientV2, ConnectConfig, ClientTelemetry, MilvusClientV2Session, and all public request, response, and type objects.

Constructor

pub async fn new(config: &ConnectConfig) -> Result<Self>

Creates a ClientV2 and connects it using the supplied ConnectConfig. The call waits up to connect_timeout for the Milvus server to become ready, and it fails with an Error when the server cannot be reached or the configuration is invalid (for example, a client certificate without a matching key).

let config = ConnectConfig::new()
    .uri("http://localhost:19530")
    .token("root:Milvus");
let client = ClientV2::new(&config).await?;

Runtime configuration

• pub fn set_rpc_deadline(&self, timeout: Duration)

  Sets the timeout applied independently to each RPC attempt.

• pub fn set_retry_param(&self, retry: RetryConfig)

  Replaces the retry policy used by subsequent RPC calls.

• pub fn telemetry(&self) -> ClientTelemetry

  Returns the shared client-side telemetry manager. See ClientTelemetry.

• pub fn session(&self, cluster_id: impl Into<String>) -> Result<MilvusClientV2Session>

  Creates a cluster-scoped session view bound to the given cluster identifier. See MilvusClientV2Session.

• pub async fn use_database(&self, database: impl Into<String>) -> Result<()>

  Selects the database used by subsequent operations on this client and its clones. See UseDatabase.

• pub fn current_database(&self) -> String

  Returns the database currently selected by this client.

• pub fn sdk_version(&self) -> &'static str

  Returns the compile-time Rust SDK package version. See SDKVersion.

Method index

Operations are grouped into the following categories. Click a category page to see the request fields and examples for each operation.

• Client — connection and runtime configuration.
• Database — database lifecycle and selection: use_database, current_database, create_database, drop_database, list_databases, describe_database, alter_database_properties, drop_database_properties.
• Collections — collection lifecycle and schema: create_collection, has_collection, describe_collection, list_collections, batch_describe_collections, get_collection_stats, drop_collection, load_collection, refresh_load, release_collection, get_load_state, truncate_collection, rename_collection, describe_replicas, alter_collection_properties, drop_collection_properties, alter_collection_field_properties, drop_collection_field_properties, add_collection_field, add_collection_struct_field, drop_collection_field, add_collection_function, alter_collection_function, drop_collection_function, add_function_field, drop_function_field.
• Vector — DML and DQL: insert, upsert, delete, query, get, search, hybrid_search, query_iterator, search_iterator.
• Partitions — partition lifecycle: create_partition, drop_partition, has_partition, list_partitions, load_partitions, release_partitions, get_partition_stats.
• Index — index management: create_index, describe_index, drop_index, list_indexes, alter_index_properties, drop_index_properties.
• Authentication — users, roles, privileges, and privilege groups.
• ResourceGroup — resource-group lifecycle, node and replica transfer.
• Snapshots — snapshot backup, restore, and pinning.
• Management — maintenance, compaction, segments, server version, and health.
• CDC — change-data-capture replication: update_replicate_configuration, get_replicate_configuration, get_replicate_info, dump_messages.
• Alias — collection aliases: create_alias, drop_alias, alter_alias, describe_alias, list_aliases.

Example

use milvus::v2::prelude::*;

let config = ConnectConfig::new()
    .uri("http://localhost:19530")
    .token("root:Milvus");
let client = ClientV2::new(&config).await?;

let schema = CollectionSchema::new()
    .add_field(
        FieldSchema::new()
            .name("id")
            .data_type(DataType::Int64)
            .primary_key(true),
    )
    .add_field(
        FieldSchema::new()
            .name("vector")
            .data_type(DataType::FloatVector)
            .dimension(4),
    );
let request = CreateCollectionRequest::builder()
    .collection_name("books")
    .schema(schema)
    .build()?;
client.create_collection(request).await?;