[Feature] Docker Compose file that is compatible with Portainer

[Boerny41]

Describe the solution you'd like
Can we get a docker compose file that is compatible with container management software like portainer? The current one gives out an error.
Failed to deploy a stack: Service ml_api Building Service tasks Building unable to prepare context: path "/data/compose/64/backend" not found

[kennethjiang]

Thank you for making this feature request. If anyone can submit a PR I'll be happy to review and merge it.

[DeathCamel58]

I got this running in Portainer by manually creating the images, then using them in the docker-compose.yml.

To build the images:

1. Download and extract this repository
2. Create a tar archive containing the ml_api folder
3. Create a tar archive containing the backend folder
4. In Portainer, go to Images, then click Build a new image
5. For name, input obico-ml_api
6. Select Upload
7. Click Select file, and select the archive of the ml_api folder
8. Click Build the image
9. Wait for it to complete
10. Go back to Images -> Build a new image
11. For name, input obico-backend
12. Select Upload
13. Click Select file, and select the archive of the backend folder.
14. Click Build the image
15. Wait for it to complete

To use the images in a stack:

1. In this repository's docker compose, replace

   build:
     context: backend
     dockerfile: 'Dockerfile'
   volumes:
     - ./backend:/app
     - ./frontend:/frontend
   

   With:

   image: obico-backend:latest
   volumes:
     - obico-backend:/app
     - obico-frontend:/frontend
   

2. Replace:

   build:
     context: ml_api
   

   With:

   image: obico-ml_api:latest
   

3. Add this to the end of the compose file:

   volumes:
     obico-backend:
     obico-frontend:
   

At this point, the stack will start, but you'll get an HTTP 500 error per #1018. Opening a shell on obico-dev-web-1 and running ./manage.py site --add server_ip_address:server_port (replacing your server IP address and port) will get you to the django web UI.

[jcolladosp]

I got this running in Portainer by manually creating the images, then using them in the docker-compose.yml.

To build the images:

1. Download and extract this repository
2. Create a tar archive containing the ml_api folder
3. Create a tar archive containing the backend folder
4. In Portainer, go to Images, then click Build a new image
5. For name, input obico-ml_api
6. Select Upload
7. Click Select file, and select the archive of the ml_api folder
8. Click Build the image
9. Wait for it to complete
10. Go back to Images -> Build a new image
11. For name, input obico-backend
12. Select Upload
13. Click Select file, and select the archive of the backend folder.
14. Click Build the image
15. Wait for it to complete

To use the images in a stack:

1. In this repository's docker compose, replace

   build:
     context: backend
     dockerfile: 'Dockerfile'
   volumes:
     - ./backend:/app
     - ./frontend:/frontend
   

   With:

   image: obico-backend:latest
   volumes:
     - obico-backend:/app
     - obico-frontend:/frontend
   

2. Replace:

   build:
     context: ml_api
   

   With:

   image: obico-ml_api:latest
   

3. Add this to the end of the compose file:

   volumes:
     obico-backend:
     obico-frontend:
   

At this point, the stack will start, but you'll get an HTTP 500 error per #1018. Opening a shell on obico-dev-web-1 and running ./manage.py site --add server_ip_address:server_port (replacing your server IP address and port) will get you to the django web UI.

How do you manage to update the images when there's a new release?

[DeathCamel58]

I believe you can re-build an image with the same name, then open the stack, go to the editor tab for the stack, and there should be a button called Update the stack.

[jcolladosp]

I believe you can re-build an image with the same name, then open the stack, go to the editor tab for the stack, and there should be a button called Update the stack.

Yes, but you also need to rebuild the .tar files, isn't it?

[DeathCamel58]

Yes. You would need to rebuild the tar files. I'd imagine that it's possible to build the images using the URL option instead of the Upload option to have Portainer clone the repository, but I haven't tried that. I wouldn't like the idea of cloning the whole repository for each of the two images.

[DeathCamel58]

Filed #1055 to try to get them to upload official prebuilt container images to a registry. This will allow us to just use their latest in a stack, so that we can easily get updates.

[kennethjiang]

I will use this issue, instead of #1055 , as the main issue since there are already discussions here.

A community member previously took a stab at it. He broke the work into 2 pieces: #824 and #821. I (prematurely) merged #824 believing that #821 will be ready soon. But eventually had to revert #824 because #821 was eventually abandoned.

The Obico cloud uses Kubernetes to run a scalable clusters. Hence we expect the self-hosted community to implement this feature because we ourselves won't be the users of this feature. I hope this makes sense.

[d-mcknight]

The Obico cloud uses Kubernetes to run a scalable clusters. Hence we expect the self-hosted community to implement this feature because we ourselves won't be the users of this feature. I hope this makes sense.

Would you be open to a PR to implement an image build pipeline in GHA similar to #824 ? If yes, should PR's be based off the master branch? I didn't see these details in the contributing guide

[kennethjiang]

The Obico cloud uses Kubernetes to run a scalable clusters. Hence we expect the self-hosted community to implement this feature because we ourselves won't be the users of this feature. I hope this makes sense.

Would you be open to a PR to implement an image build pipeline in GHA similar to #824 ? If yes, should PR's be based off the master branch? I didn't see these details in the contributing guide

Yes please send PR. Just make sure you test it before you set the PR to be ready for review. Also I just updated the contributing guidelines. It may still not be clear enough. Feel free to ask if it's the case.

[d-mcknight]

Looks good to me; I got a decent start on this last night and will open a draft PR later today once I confirm at least minimum functionality is there