'Index' object has no attribute 'configuration' when running my LangChain application in a docker image

Question

This is my code:

from langchain_community.vectorstores import Pinecone
from langchain_openai import ChatOpenAI
from langchain_openai import OpenAIEmbeddings
from langchain.chains import ConversationalRetrievalChain

pc = pinecone.Pinecone(api_key=secret['PINECONE_API_KEY'], 
environment=secret['PINECONE_ENV']) 

index = pc.Index(secret['PINECONE_INDEX_NAME'])
embeddings = OpenAIEmbeddings(secret['OPENAI_API_KEY'])
model = ChatOpenAI(model_name='gpt-4-turbo-preview')

docsearch = Pinecone.from_existing_index(index_name=secret['PINECONE_INDEX_NAME'], embedding=embeddings, namespace=secret['PINECONE_NAMESPACE']), search_kwargs = {'k': 25, 'namespace': secret['PINECONE_NAMESPACE']}
retriever = docsearch.as_retriever(namespace=secret['PINECONE_NAMESPACE'], search_kwargs=search_kwargs)

qa = ConversationalRetrievalChain.from_llm(llm=model,retriever=retriever)
qa({'question': prompt, 'chat_history': chat})
    

When I run this locally, I get a proper response. When I run it through docker, I get 'Index' object has no attribute 'configuration'.

These are the versions I'm running:

langchain==0.1.7
langchain-community==0.0.20
pinecone-client==3.0.2

Any thoughts? This has been wrecking my brain all morning