dump_messages()

Adds message dumping for data-salvage workflows. Async variant shares the sync method contract.

Request Syntax

def dump_messages( self, pchannel: str, start_message_id: Dict, start_timetick: int = 0, end_timetick: int = 0, timeout: Optional[float] = None, **kwargs, ):

PARAMETERS:

• pchannel (str) - [REQUIRED] The name of the physical channel from which to dump messages.

• start_message_id (Dict) - [REQUIRED] The starting write-ahead-log position, expressed as an object with id and wal_name. The message_id from get_replicate_info() can be passed directly.

• start_timetick (int) - Default: 0 The inclusive lower timetick bound. Use 0 to omit the lower bound.

• end_timetick (int) - Default: 0 The inclusive upper timetick bound. Use 0 to keep streaming until the RPC is cancelled.

• timeout (Optional[float]) - Default: None The maximum time, in seconds, allowed for the entire stream.

• kwargs (Any) - The additional keyword arguments passed to the underlying client or request.

RETURN TYPE:

Generator[dict, None, None]

RETURNS:

Generator yielding dictionaries with message_id, payload, and properties for each WAL message.

EXCEPTIONS:

• MilvusException Raised when the server rejects the request or the RPC fails. Inspect the server error message for exact failure details.

Examples

Demonstrates dump messages usage.

from pymilvus import MilvusClient

client = MilvusClient(uri="http://localhost:19530", token="root:Milvus")
for message in client.dump_messages(pchannel="by-dev-rootcoord-dml_0", start_message_id={"id": "message-id", "wal_name": "default"}, start_timetick=0, timeout=30):
    print(message)