# Run Discourse as docker swarm service

**URL:** https://meta.discourse.org/t/run-discourse-as-docker-swarm-service/352357
**Category:** Self-hosting
**Created:** [February 14, 2025, 7:07pm UTC](https://meta.discourse.org/t/run-discourse-as-docker-swarm-service/352357 "2025-02-14T19:07:12Z")
**Posts on this page:** 2
**Page:** 1

<div class="post-metadata">

### Author: ![tsr1](https://avatars.discourse-cdn.com/v4/letter/t/f19dbf/32.png) [@tsr1](https://meta.discourse.org/u/tsr1)
#### Post date: [February 14, 2025, 7:07pm UTC](https://meta.discourse.org/t/run-discourse-as-docker-swarm-service/352357/1 "2025-02-14T19:07:12Z")

</div>

I am attempting to deploy Discourse as a docker swarm service.  
Used the official launcher method to deploy Discourse in a single-container setup.  
Copied the required files from /var/www/discourse inside the standalone container to a location on the host servers to use in the Swarm service.  
Used an existing postgre server  
Created a dedicated Redis container for Discourse  
Discourse fails to start and exits with:  
uninitialized constant Discourse::SiteSetting (NameError)  
The container connects to PostgreSQL and Redis without issues.  
DISCOURSE\_HOSTNAME is correctly set to discourse.xyz in the environment.  
Could this be due to missing dependencies or a broken asset compilation?  
Does Discourse require any modifications to properly function within a Swarm environment?

```plaintext
version: '3.7'

services:
  app:
    image: discourse-app:latest
    user: "1000:1000"
    working_dir: /var/www/discourse
    deploy:
      replicas: 1
      restart_policy:
        condition: on-failure
      labels:
        - "traefik.enable=true"
        - "traefik.backend=discourse"
        - "traefik.docker.network=service"
        - "traefik.frontend.rule=Host:discourse.xyz"
        - "traefik.port=80"
    command:
      - "/bin/bash"
      - "-c"
      - |
        bundle config set deployment true && \
        bundle config set without 'test development' && \
        RAILS_ENV=production bundle install && \
        RAILS_ENV=production bundle exec rake db:migrate && \
        RAILS_ENV=production bundle exec rails server -b 0.0.0.0
    environment:
      DISCOURSE_DB_HOST: "postgrest1"
      DISCOURSE_DB_PORT: "5432"
      DISCOURSE_DB_USERNAME: "postgres"
      DISCOURSE_DB_PASSWORD: " ****"
      DISCOURSE_DB_NAME: "discourse"
      DISCOURSE_REDIS_HOST: "redis-discourse"
      DISCOURSE_REDIS_PORT: "6379"
      DISCOURSE_REDIS_PASSWORD: " ****"
      DISCOURSE_HOSTNAME: "discourse.xyz"
      DISCOURSE_DEVELOPER_EMAILS: "abc@xyz.com"
      DISCOURSE_SMTP_ADDRESS: "xyz"
      DISCOURSE_SMTP_PORT: "25"
      DISCOURSE_NOTIFICATION_EMAIL: "donotreply@xyz.com"
    networks:
      - service
    ports:
      - "8080:80"
      - "8443:443"
    volumes:
      - type: bind
        source: /discourse/discourse_files/discourse
        target: /var/www/discourse
        bind:
          propagation: rprivate

      - type: bind
        source: /discourse/shared/standalone
        target: /shared
        bind:
          propagation: rprivate

      - type: bind
        source: /discourse/shared/standalone/log/rails
        target: /data/modulusdata/infra/discourse/shared/standalone/log/rails
        bind:
          propagation: rprivate

      - type: bind
        source: /discourse/shared/standalone/log/var-log
        target: /var/log
        bind:
          propagation: rprivate

networks:
  service:
    external: true

```

---

<div class="post-metadata">

### Author: ![pfaffman](https://sea3.discourse-cdn.com/meta/user_avatar/meta.discourse.org/pfaffman/32/120154_2.png) [@pfaffman](https://meta.discourse.org/u/pfaffman)
#### Post date: [February 14, 2025, 8:43pm UTC](https://meta.discourse.org/t/run-discourse-as-docker-swarm-service/352357/2 "2025-02-14T20:43:34Z")

</div>

This is beyond the support you can expect here. There are just too many tiny things that could be wrong that’s why the [standard install](https://meta.discourse.org/t/142537?silent=true) is really all that is supported. Here are some hints. If you’re stuck and have a budget you can contact me or ask in #Marketplace

You can’t use the base image. You have to bootstrap your own and push it to a repo. If you’re clever and have lots of time you can get github to do it for you.

You’ll need to put your assets on S3 ([Configure an S3 compatible object storage provider for uploads](https://meta.discourse.org/t/configure-an-s3-compatible-object-storage-provider-for-uploads/148916))

Are you trying to have all of those write to the same log files? I’d just leave them in the container and ignore them.

I don’t think you need ports 8080 and 8443—traefik is handling that, right? Discourse won’t work on a non-standard port.

something like:

```plaintext
./launcher bootstrap app
docker push-that-container-to-your-private-repo
./launcher start-cmd

```

Then you can use the variables in the start-cmd to launch your swarm with the right stuff so it can find your database, redis, and so on. I see you have a bunch, but I’m not sure that’s all that you need.