ok. thanks for the fast answer.

@mdc773 there is a video at https://forum.cloudron.io/topic/2843/read-first-before-starting-to-package-an-app which @fbartels made

Thanks for the reminder, I have updated that file now with the new base image.

I think that is just the docker image and not the docker file - https://hub.docker.com/r/jacobmatrktippetts/monero-node

Maybe a temporary issue with your private docker registry or otherwise. Lets see if next time we have more info to work with.

I solved the port issue as well. The problem occured when I tried to start an interactive shell, which counted as starting a second application inside the cloudron app, that's where the port conflict came from. If someone else is working with Elixir on Cloudron, you can start an interactive shell like this: /app/_build/prod/rel/APP/bin/APP remote

@alex-uxlabsmx should work now. there were a couple of issues in the crontab: '&' in the end of the cron line. this puts the command in the background. cronjobs are not run in background and instead run on a schedule. so, the '&' should be removed. 'source' command is a bash primitive. you have to wrap it in bash -c "source ... && anothercommand", for example seems to work after that.

I am not familiar with how nestjs works during startup, but something tries to wholesale rm the /dist folder. I guess it tries to rebuild the app on startup? You may have to copy the app from read-only to /run first thing in start.sh and then start the app and not symlink the folder? Just a guess though.

Glad you got it sorted out!

Thank you for the detailed response! Since this app doesn't run the LLM itself, the system requirements are less heavy. But wouldn't everything that wasn't attached to Cloudron just run inside the container? I mean, let's say I write a dockerfile for this app, but "forget" to use the postgres addon of cloudron. Wouldn't the app still run, but with the database inside the app, so it would be overwritten upon updating? Other than that, this app seems to have a sandbox container and one for vector databases, which couldn't be rebuilt easily in Cloudron.

@girish To make it more clear, I have made a template repository with all the files and scripts I usually use for my Cloudron apps. In the future, we might benefit from adopting such a strategy for the Cloudron CLI init script by using these files as a template/starter Cloudron app project. It'd make maintenance easier, more uniform, and make the process less daunting for new app devs. https://github.com/Lanhild/starter-cloudron

@Lanhild yup, we already started using multi-stage builds by now. For app store, you have to pin the images though. Using ":latest" is not good will have to be changed.

Got it. Thanks for the answer!

@girish scroll and copy > chatgpt

There‘s a whole category for that: https://forum.cloudron.io/category/5/app-wishlist I‘d suggest creating one topic per app. If your app generates enough interest, some kind soul might gift you (and us) with their spare time

It would reach it via the public domain I would guess, so it is not dependent on the local docker network addresses. All this is not an elegant solution for sure but to have this well supported we would need to have a quite sophisticated reverseproxy/nginx config UI somehow.

i have to test it in production so i've done it here : https://git.cloudron.io/nils/pocketbase i may improve this package ,handling Cloudron user management and email configuration if people or i find interest in it .