Flush()

This operation flushes the streaming data and seals segments. It is recommended to call this operation after all the data has been inserted into a collection.

Status Flush(const FlushRequest& request)

Request Syntax

auto request = FlushRequest()
    .WithDatabaseName(db_name)
    .WithCollectionNames(names)
    .WithWaitFlushedMs(ms);

REQUEST METHODS:

• WithDatabaseName(const std::string& db_name)

  Sets the target database name. The default database applies if it is empty.

• WithCollectionNames(std::set<std::string>&& names)

  Sets the name of the collection.

• AddCollectionName(const std::string& name)

  Adds the name of a collection to flush.

• WithWaitFlushedMs(int64_t ms)

  Sets the number of milliseconds to wait for the flush action to complete. The default value is 0. If WaitFlushedMs is set to 0, this operation repeatedly calls GetFlushState() to check the status of related segments until all segments are flushed, ensuring that the buffer is persisted successfully. If WaitFlushedMs is greater than 0, this operation will break the loop after a specified period of time and return a status indicating a timeout.

RETURNS:

Status

Check status.IsOk() to confirm success.

EXCEPTIONS:

• StatusCode

  Check status.Code() and status.Message() for error details.

Example

#include "milvus/MilvusClientV2.h"
auto client = milvus::MilvusClientV2::Create();

milvus::ConnectParam connect_param{"http://localhost:19530", "root:Milvus"};
auto status = client->Connect(connect_param);
if (!status.IsOk()) {
    std::cout << status.Message() << std::endl;
}

// call flush() here just to persist the data so that indexnode can build index on a new segment
// Note: in practice, no need to call flush() manually since milvus automatically trigger flush actions
status = client->Flush(milvus::FlushRequest().AddCollectionName(collection_name));
if (!status.IsOk()) {
    std::cout << status.Message() << std::endl;
}