Issues with running n8n with ollama locally

[Soroush.M99]

Hello everyone so I did everything needed for running n8n with ollama locally (installing ollama, downloading some models, n8n ai starter kit, installing docker, …) I accessed n8n locally with http://localhost:5678/ as guided, and here’s the docker setting, but when I want to fetch the agent with the ollama model (with the base URL http://localhost:11434, also as guided), it doesn’t connect, and I uploaded the errors on both n8n’s and docker’s sides.

When I click on n8n to connect with Docker, it errors:
The service refused the connection - perhaps it is offline. n8n | connect ECONNREFUSED ::1:11434

Can you guys help me on this? I’ll be appreciated. Been thinking of just breaking this PC for two days lol.

[Franz]

services:
  database:
    image: postgis/postgis:13-master
    restart: always
    ports:
      - 5432:5432
    volumes:
      - ./postgres/data:/var/lib/postgresql/data
    environment:
      POSTGRES_USER: ${POSTGRES_USER}
      POSTGRES_PASSWORD: ${POSTGRES_PASSWORD}
      POSTGRES_DB: "n8n"
    healthcheck:
      test: [ 'CMD-SHELL', 'pg_isready -h localhost -U ${POSTGRES_USER} -d n8n' ]
      interval: 5s
      timeout: 5s
      retries: 10

  n8n:
    image: n8n
    restart: always
    build:
      dockerfile: n8n/Dockerfile
    ports:
      - 5678:5678
    volumes:
      - ./n8n/data:/home/node/.n8n
      - ./n8n/credentials:/home/node/credentials
      - ./n8n/workflows:/home/node/workflows
    environment:
      DB_POSTGRESDB_DATABASE: "n8n"
      DB_POSTGRESDB_HOST: "database"
      DB_POSTGRESDB_PASSWORD: ${POSTGRES_PASSWORD}
      DB_POSTGRESDB_PORT: "5432"
      DB_POSTGRESDB_USER: ${POSTGRES_USER}
      DB_TYPE: "postgresdb"
      DOMAIN_NAME: "n8n.example.com"
      EXECUTIONS_DATA_MAX_AGE: 168
      EXECUTIONS_DATA_PRUNE: true
      EXECUTIONS_DATA_PRUNE_MAX_COUNT: 50000
      EXECUTIONS_DATA_SAVE_MANUAL_EXECUTIONS: false
      EXECUTIONS_DATA_SAVE_ON_ERROR: "none"
      EXECUTIONS_DATA_SAVE_ON_PROGRESS: false
      EXECUTIONS_DATA_SAVE_ON_SUCCESS: "none"
      GENERIC_TIMEZONE: "Europe/Vienna"
      N8N_ENFORCE_SETTINGS_FILE_PERMISSIONS: true
      N8N_LOG_LEVEL: "error"
      NODE_FUNCTION_ALLOW_BUILTIN: "*"
      NODE_FUNCTION_ALLOW_EXTERNAL: "*"
      TZ: "Europe/Vienna"
      WEBHOOK_URL: "https://n8n.example.com"
      N8N_ENCRYPTION_KEY: ${N8N_ENCRYPTION_KEY}
    links:
      - database
      - redis
    depends_on:
      redis:
        condition: service_healthy
      database:
        condition: service_healthy

  redis:
    image: redis:latest
    restart: always
    ports:
      - "6379:6379"
    environment:
      REDIS_PASSWORD: ${REDIS_PASSWORD}
      REDIS_PORT: 6379
      REDIS_DATABASES: 16
    volumes:
      - ./redis/data:/data
      - ./redis/redis.conf:/usr/local/etc/redis/redis.conf
    command: ["redis-server", "/usr/local/etc/redis/redis.conf"]
    healthcheck:
      test: ['CMD', 'redis-cli', 'ping']
      interval: 5s
      timeout: 5s
      retries: 10

  ollama:
    image: ollama/ollama:latest
    restart: always
    ports:
      - "11434:11434"
    volumes:
      - ./ollama:/root/.ollama
    environment:
      - OLLAMA_MODELS=/root/.ollama/models

How do you start the container?

The ports are only available to your host but not to each other.

A solution using Docker Compose would be relatively simple.

All containers would then run in a network.

You can access other containers using the service name (e.g., ollama:11434) instead of localhost:11434.

Without docker-compose you have to create a network first
docker network create example-network
Then you have to start the containers with the network.
docker run -d --name ollama --network example-network -p 11434:11434 ollama/ollama:latest
docker run -d --network example-network http://docker.n8n.io/n8nio/n8n
Then you can access ollama from your n8n container with ollama:11434.

[Soroush.M99]

THANK YOU SO MUCH SIR IT’S NOW CONNECTED REALLY APPRECIATED. But the only thing is actually I downloaded 3 models for ollama but I couldnt find and select it is the n8n setup for the model. There’s only Ollama 3.2. Can you please help me on that too? tnx
Also, since I’m super amateur, I just ran those three lines at the end of your answer in the Docker terminal, and it worked and connected just fine. Maybe I didn’t follow your instructions completely somewhere because I wasn’t very skilled.

tnx again.

[Franz]

Did you download the 3 models locally on Windows instead of in the container?

Please try running docker exec -it ollama ollama list. The first “ollama” is the name of your container.

The screenshot briefly explains how to find the container name (mine is n8n because I don’t have ollama running).

[Soroush.M99]

Hello and THANK YOU AGAIN, SIR.
Thanks to your help, I was able to realize I didn’t download the models on the container, so with this command on the docker, I downloaded one of them:
docker exec ollama ollama pull gemma3:4b
When I got back to N8N, fortunately a new model was added, but I don’t know why it doesn’t respond to my input.
I’m really sorry I bother you a lot. I’ll be very appreciative if you can help me also with this. Thanks again.

edit: i added ‘deepseek-r1:8b’ to the models in docker again and it works just fine in n8n local this time idk what is it’s problem with gemma3.