Text2vec-ollama module ignores OLLAMA_API_ENDPOINT and defaults to localhost in Weaviate v1.31.2 on Linux

General
1

I’m encountering a persistent network issue when trying to use the text2vec-ollama module with a locally running Ollama instance, and I would be very grateful for some guidance.

My Goal: To have a Weaviate container (v1.31.2) use its text2vec-ollama module to connect to an Ollama service running on the host machine (Ubuntu).

The Problem: When I try to insert data into a collection that uses the text2vec-ollama vectorizer, the operation fails with a connection refused error. The error log clearly shows that Weaviate is trying to connect to http://localhost:11434, even though I have explicitly set the OLLAMA_API_ENDPOINT environment variable to http://host.docker.internal:11434.

1. Environment Details

  • Weaviate Version: 1.31.2 (from semitechnologies/weaviate:1.31.2 image)
  • Weaviate Client: Python (weaviate-client, 4.15.2)
  • Docker Environment: Docker on Ubuntu Linux(24.04.2)
  • Ollama Service: Running successfully on the host machine, accessible at http://localhost:11434 from the host’s terminal.

2. My docker-compose.yml Configuration

This is the configuration I am currently using, which includes various attempts to solve the issue.

services:
  weaviate:
    image: semitechnologies/weaviate:1.31.2
    ports:
      - "8080:8080"
      - "50051:50051"
    volumes:
      - ./weaviate_data:/var/lib/weaviate
    restart: on-failure:0
    extra_hosts:
      - "host.docker.internal:host-gateway"
    environment:
      QUERY_DEFAULTS_LIMIT: 25
      AUTHENTICATION_ANONYMOUS_ACCESS_ENABLED: 'true'
      PERSISTENCE_DATA_PATH: '/var/lib/weaviate'
      DEFAULT_VECTORIZER_MODULE: 'none'
      CLUSTER_HOSTNAME: 'node1'
      ENABLE_TOKENIZER_GSE: 'true'
      ENABLE_MODULES: 'text2vec-ollama,generative-ollama'
      OLLAMA_API_ENDPOINT: 'http://host.docker.internal:11434'

3. Observed Behavior & Error Message

When I run my Python script to insert data, the batch insertion fails for all objects. The error message is consistently:

ERROR:weaviate-client:{'message': 'Failed to send all objects in a batch...', 'error': 'WeaviateInsertManyAllFailedError(\'Every object failed during insertion. Here is the set of all errors: send POST request: Post "http://localhost:11434/api/embeddings": dial tcp [::1]:11434: connect: connection refused\')'}

This confirms the module is trying to connect to localhost inside the container, not host.docker.internal.

4. Diagnostic Steps Already Taken

To isolate the problem, I have performed the following tests from inside the Weaviate container (docker compose exec weaviate /bin/sh):

  • Test 1: Internet Connectivity: ping -c 4 8.8.8.8 → SUCCESS. The container can reach the internet.
  • Test 2: Host DNS Resolution: curl http://host.docker.internal:11434 → FAILED with Could not resolve host: host.docker.internal. This indicates that on my Linux system, this special DNS name is not resolving correctly.
  • Test 3: Collection Meta Info Check: Print the collection meta infomation, and two modules generative-ollamaandtext2vec-ollama have been loaded successfully.
{'grpcMaxMessageSize': 104858000, 'hostname': 'http://[::]:8080', 'modules': {'generative-ollama': {'documentationHref': 'https://github.com/ollama/ollama/blob/main/docs/api.md#generate-a-completion', 'name': 'Generative Search - Ollama'}, 'text2vec-ollama': {'documentationHref': 'https://github.com/ollama/ollama/blob/main/docs/api.md#generate-embeddings', 'name': 'Ollama Module'}}, 'version': '1.31.2'}

The only workaround I’ve found is to set network_mode: "host" for the Weaviate service. While this works for local development, it’s not an ideal solution as it breaks network isolation.

5. A Tiny Python Test Case

import weaviate
import time
from weaviate.classes.config import Configure, Property, DataType, Tokenization
from weaviate.exceptions import WeaviateQueryException

# --- settings ---
COLLECTION_NAME = "KeywordSearchDiagnosticTest"
TEST_SENTENCE = "Weaviate是一个开源的向量数据库，非常适合RAG系统。"

def run_diagnostic():
    client = None
    try:
        # --- step 1: connect to Weaviate and delete old test collection ---
        print("--- step 1: connect to Weaviate and delete old test collection ---")
        client = weaviate.connect_to_local()
        if client.collections.exists(COLLECTION_NAME):
            client.collections.delete(COLLECTION_NAME)
            print(f"✅ delete the old test collection: '{COLLECTION_NAME}'。")
        else:
            print(f"✅ old test collection doesn't exit。")

        # --- step2: create a new Collection ---
        print("\n--- tep2: create a new Collection ---")
        client.collections.create(
            name=COLLECTION_NAME,
            properties=[
                Property(
                    name="content",
                    data_type=DataType.TEXT,
                    tokenization=Tokenization.GSE,  # set tokenization to GSE
                    index_searchable=True          
                )
            ],
            vectorizer_config=[Configure.NamedVectors.text2vec_ollama(
                name="content_vector",
                source_properties=["content"],
                model="bge-m3:latest"
            )]
        )
        print(f"✅ collection '{COLLECTION_NAME}' has been created successfully。")

        # --- step 3: add a test info ---
        print("\n--- step 3: add a test info ---")
        collection = client.collections.get(COLLECTION_NAME)
        uuid = collection.data.insert({
            "content": TEST_SENTENCE
        })
        print(f"✅ add test info successfully，UUID: {uuid}...")

    except Exception as e:
        print(f"\n❌ meet issue or error: {e}")
    finally:
        if 'client' in locals() and client.is_connected():
            client.close()
            print("\n task completed, close the connection。")

if __name__ == "__main__":
    run_diagnostic()

You need to dowload ollama and the embedding model before you run the above code