what cause Duplicated index ?

[run-goto]

When I delete an embedded index， I get a warning that collection have a duplicate index。

[yhmo]

Could you offer a snippet to reproduce the problem?

create_index/drop_index work well in my script:

import random

from pymilvus import (
    MilvusClient, DataType, 
)

client = MilvusClient(uri="http://localhost:19530")
print(client.get_server_version())

collection_name = "AAA"

schema = MilvusClient.create_schema()
schema.add_field(field_name="id", datatype=DataType.INT64, is_primary=True, auto_id=False)
schema.add_field(field_name="embedding", datatype=DataType.FLOAT_VECTOR, dim=768)


client.drop_collection(collection_name=collection_name)
client.create_collection(
    collection_name=collection_name,
    schema=schema,
)
print(collection_name, "created")

index_params = client.prepare_index_params()
index_params.add_index(
    field_name="embedding",
    index_name="embedding",
    index_type="AUTOINDEX",
    metric_type="IP",
)
client.create_index(collection_name=collection_name, index_params=index_params)
print("index created")

client.drop_index(collection_name=collection_name, index_name="embedding")
print("index dropped")

[yhmo]

It allows repeatedly calling create_index() with the same parameters, actually only one index is created on the field. Your script doesn't throw any errors.

[run-goto]

but when i delete index, I get a warning that collection have a duplicate index。

[yhmo]

I added three lines under your script to delete index. I didn't get the warning.
My milvus is v2.5.11.

import threading

from pymilvus import (
    MilvusClient, DataType,
)


def create_collection(collection_name):
    client = MilvusClient(uri="http://localhost:19530")
    print(client.get_server_version())

    schema = MilvusClient.create_schema()
    schema.add_field(field_name="id", datatype=DataType.INT64, is_primary=True, auto_id=False)
    schema.add_field(field_name="embedding", datatype=DataType.FLOAT_VECTOR, dim=768)

    index_params = client.prepare_index_params()
    index_params.add_index(
        field_name="embedding",
        index_name="embedding",
        index_type="AUTOINDEX",
        metric_type="IP",
    )
    client.create_collection(
        collection_name=collection_name,
        schema=schema,
        index_params=index_params
    )
    print(collection_name, "created")

    client.create_index(collection_name=collection_name, index_params=index_params)
    print("index01 created")


# 启动多个线程创建不同集合
threads = []
for i in range(10):
    collection_name = "CCC"
    thread = threading.Thread(target=create_collection, args=(collection_name,))
    threads.append(thread)
    thread.start()

for thread in threads:
    thread.join()

# delete index
client = MilvusClient(uri="http://localhost:19530")
client.release_collection(collection_name="CCC")
client.drop_index(collection_name="CCC", index_name="embedding")

[run-goto]

my milvus is v2.5.4. you can try this version. if having two same indexes will affect query efficiency?

[yhmo]

Yes, I can reproduce it on v2.5.4.
Once the duplicated indexes are created, it is not possible to drop it even I upgrade the milvus to v2.5.11. I just raised an issue to track this bug: #42055
I think the higher version milvus should be able to drop the duplicated indexes.