Data validation and `upload_collection` method

[rushter]

I think the upload_collection method should validate data before uploading it.

For example, it will silently upload 1 million records where vector_size=256 to a collection where vector_size=512. As a result, you get no errors and no data will be added to a collection.

Comparing the vector size of a collection to the input array size is very cheap.

[arestifo]

Just ran into the silent fail issue, I'm surprised the Python client doesn't report an error here. I see a warning in the Qdrant Docker logs but nothing is wrong from the Python side: [2023-05-08T21:58:47.845Z WARN collection::collection_manager::collection_updater] Update operation declined: Wrong input: Vector inserting error: expected dim: 128, got 1536

[generall]

it might happen if you use wait=false in upsert API

[zeynepiskndr]

Just ran into the silent fail issue, I'm surprised the Python client doesn't report an error here. I see a warning in the Qdrant Docker logs but nothing is wrong from the Python side: [2023-05-08T21:58:47.845Z WARN collection::collection_manager::collection_updater] Update operation declined: Wrong input: Vector inserting error: expected dim: 128, got 1536

I have the same problem, I would appreciate it if you could help. I am working with the Google Cloud Server and I came across this error in Notepad. UnexpectedResponse: Unexpected Response: 400 (Bad Request) Raw response content: b'{"status":{"error":"Wrong input: Vector inserting error: expected dim: 734, got 0"},"time":0.002494573}'

[kientv]

I'm hitting this error: grpc_message:"Wrong input: Vector inserting error: expected dim: 1536, got 768"

[joein]

I'm hitting this error: grpc_message:"Wrong input: Vector inserting error: expected dim: 1536, got 768"

it means that you are trying to upload 768-d vectors to a collection which has been configured to work with 1536-d vectors

[AI-Aether]

Is there a fix? Even setting wait=True is not solving this issue

[kientv]

That's not Qdrant's problem. You can solve that error by deleting old collection and create a new collection with correct dimension size. for instance: OpenAI embedding is 1536, Gemini is 768.