Does the Python milvus SDK support asynchronous queries? If supported, how should it be written? Can you provide an example? Thank you very much.

[bestxiaoxiaoming-hm]

我在使用_async=true的时候，无法从SearchFuture取出结果

[yhmo]

MilvusClient doesn't support future results.
The Collection class supports it. https://github.com/milvus-io/pymilvus/blob/da51ba14edde3c67b682f2d0ae1b85a095815692/pymilvus/orm/collection.py#L816

from pymilvus import (
    FieldSchema, CollectionSchema, DataType,
    utility, Collection, connections
)
import random
connections.connect("default", host="localhost", port=19530)

dim = 128
schema = CollectionSchema(fields=[
    FieldSchema(name="id", dtype=DataType.INT64, is_primary=True, auto_id=False),
    FieldSchema(name="vector", dtype=DataType.FLOAT_VECTOR, dim=dim),
])

collection_name = "AAA"

utility.drop_collection(collection_name)
collection = Collection(collection_name, schema)
print("collection created")

batch = 1000
collection.insert(data=[
    [i for i in range(batch)],
    [[random.random() for _ in range(dim)] for _ in range(batch)],
])


index_params_ivf={"index_type": "FLAT", "params": {}, "metric_type": "L2"}
collection.create_index("vector", index_params=index_params_ivf)
collection.load()

count = collection.query(expr="", output_fields=["count(*)"], consistency_level = "Strong")
print(count)


future_res = collection.search(data=[[random.random() for _ in range(dim)]],
                            anns_field="vector",
                            limit=10,
                            param={},
                            _async=True,
                            output_fields=["*"])
# do something other
print("=================")

# get the result
results = future_res.result()
for hits in results:
    for hit in hits:
        print(hit)

[lingvisa]

In Milvus 2.5, does MilvusClient support asynchronous queries? I want to see whether I should use Collection if MilvusClient doesn't support async yet.

[yhmo]

In the pymilvus v2.5.3, we provide AsyncMilvusClient for this purpose: https://github.com/milvus-io/pymilvus/blob/v2.5.3/pymilvus/milvus_client/async_milvus_client.py

AsyncMilvusClient uses async/await, you can read the examples to know how to use it:
https://github.com/milvus-io/pymilvus/blob/v2.5.3/examples/simple_async.py
https://github.com/milvus-io/pymilvus/blob/v2.5.3/examples/simple_async2.py

[lingvisa]

That's very cool! By default, AsyncMilvusClient should be used. right?

[yhmo]

Both optional.
If you want sync call, use the MilvusClient:

milvus_client = MilvusClient(uri=uri)
milvus_client.

If you want async call, use AsyncMilvusClient

milvus_client = AsyncMilvusClient(uri=uri)