BulkImport

The BulkImport struct is an async client for the Milvus bulk-import REST endpoints. Create one with BulkImport::new(), then use its methods to create import jobs, list jobs, and monitor progress. The create operation is also called bulk import.

pub struct BulkImport

Constructor

Constructs a bulk-import REST client from the given configuration.

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

PARAMETERS:

• config: &BulkImportConfig

  Connection settings for the bulk-import REST client. The url must be a valid http or https URL containing a host and no query string or fragment. TLS-related fields are validated together.

RETURNS:

Result

Returns a BulkImport client on success, or Error when the configuration is invalid.

BulkImportConfig

Connection settings for the bulk-import REST client.

pub struct BulkImportConfig

PARAMETERS:

• url: String

  Base URL of the Milvus bulk-import REST endpoint. Required.

• api_key: String

  API key sent as a Bearer token in the Authorization header.

• timeout: Duration

  Overall timeout for each HTTP request. A zero duration disables the client-side request timeout. Defaults to 20 seconds.

• verify_tls: bool

  Whether to verify the server’s TLS certificate. Defaults to true.

• ca_certificate_path: Option<PathBuf>

  Path to a PEM CA certificate used to verify the server. Cannot be combined with disabled TLS verification.

• client_identity_path: Option<PathBuf>

  Path to one PEM file containing the client certificate and private key. Cannot be combined with separate client certificate or private-key paths.

• client_certificate_path: Option<PathBuf>

  Path to a PEM client certificate, used together with client_private_key_path.

• client_private_key_path: Option<PathBuf>

  Path to a PEM client private key, used together with client_certificate_path.

Methods

BulkImport()

Creates a bulk-import job.

pub async fn bulk_import(&self, request: BulkImportRequest) -> Result<BulkImportResponse>

The request must specify exactly one data source among files, object_url, object_urls, and data_paths.

REQUEST FIELDS:

• database_name: String

  Name of the database that holds the collection.

• collection_name: String

  Name of the target collection. Required.

• partition_name: String

  Name of the target partition.

• files: Vec<Vec<String>>

  Local or object-storage paths visible to an open-source Milvus deployment, grouped per import.

• object_url: String

  Deprecated singular object URL accepted by Milvus 2.6. Prefer object_urls.

• object_urls: Vec<Vec<String>>

  Object URLs of the files to import, grouped per import.

• cluster_id: String

  Cluster ID for Zilliz Cloud deployments. Cannot be combined with project_id and region_id.

• project_id: String

  Project ID for Zilliz Cloud deployments. Must be specified together with region_id.

• region_id: String

  Region ID for Zilliz Cloud deployments. Must be specified together with project_id.

• access_key: String

  Object-storage access key. Must be specified together with secret_key.

• secret_key: String

  Object-storage secret key. Must be specified together with access_key.

• token: String

  Object-storage session token.

• volume_name: String

  Volume name, required when data_paths is used.

• data_paths: Vec<Vec<String>>

  Data paths within the volume to import, grouped per import.

• options: HashMap<String, Value>

  Additional import options, such as auto_commit.

RETURNS:

Result

BulkImportResponse exposes code(), message(), data(), and the convenience accessors job_id(), state(), progress(), and reason(). Returns Error on failure.

Example

let request = BulkImportRequest::builder()
    .database_name("books_db")
    .collection_name("books")
    .file("one.parquet")
    .file_group(["id.npy", "embedding.npy"])
    .option("auto_commit", serde_json::json!(false))
    .build()?;
let resp = bulk.bulk_import(request).await?;
println!("job_id: {:?}", resp.job_id());

Example

let bulk = BulkImport::new(
    &BulkImportConfig::new()
        .url("http://localhost:19530")
        .api_key("root:Milvus"),
)?;