GRPC Resource Exhausted Error

Support
1

ErrorObject(message=’WeaviateBatchError(\‘Query call with protocol GRPC batch failed with message <AioRpcError of RPC that terminated with:\n\tstatus = StatusCode.RESOURCE_EXHAUSTED\n\tdetails = “CLIENT: Sent message larger than max (146965530 vs. 104858000)“\n\tdebug_error_string = “UNKNOWN:Error received from peer {created_time:“2024-08-22T13:03:51.442771-07:00”, grpc_status:8, grpc_message:“CLIENT: Sent message larger than max (146965530 vs. 104858000)“}”\n>.\‘)’,

I am trying to vectorize images from Google Cloud Platform with Weaviate but am getting RESOURCE_EXHAUSTED errors. I am using the multimodal model for this. When I was testing this on the Sandbox it worked completely fine, but now it is not working on the serverless cluster. Is it that the image sizes are too big (confusing because it worked on the Sandbox) and if so, how should I workaround this? I tried to vectorize 100 images: 72 worked 28 failed with that error above.

I am using 1.25.10 and the code is below:

image_b64 = url_to_base64(src_obj["image_url"])
        weaviate_obj = {
            "image": image_b64,
            "image_url": src_obj["image_url"],
            "image_name": src_obj["image_name"],
            "source": src_obj["source"],
        }
        obj_uuid = src_obj["uuid"] if src_obj["uuid"] else generate_uuid5(weaviate_obj)

        batch.add_object(
            properties=weaviate_obj,
            uuid = obj_uuid
            # vector=vector  # Optionally provide a pre-obtained vector
        )
2

hi @Cobyboss !

For all clusters hosted in our cloud, the best place for support is opening a support ticket. Check the Support Link inside your console dashboard for details.

This error message indicates that the size of the payload was too big to be handled.

How are you configuring your batch size? What batch sizes have you tried?

here is an example:

with collection.batch.fixed_size(batch_size=10, concurrent_requests=2) as batch:
    for data_row in objects:
        batch.add_object(
            properties=data_row,
            uuid=generate_uuid5(data_row.get("reference_id"))
        )

Let me know if this helps!