Cloudron Forum

@shanelord01 thanks for bringing this up Shane. I would like to see it packaged, it will probably need to start as a custom app.

I installed one myself taking over a full KVM-1 VPS at Hostinger ($4/month).

I'm an AT Protocol expert and maintain a tech talks Ghost blog at https://atprotocol.dev (hosted on my Cloudron server!).

There are some misunderstandings in this thread about architecture and frankly I'm not going to argue about decentralization. Running a PDS means user ownership of your data, which seems very on mission for Cloudron.

It also handles auth, acting as your OAuth server wherever you log into with your account. There are multiple non-Bluesky apps with different data types (called Lexicons). A few of my favourites:

Smoke Signal - events & RSVP https://smokesignal.events White Wind - markdown blogs https://whtwnd.com Frontpage - link ranking & discussion https://frontpage.fyi Recipe Exchange - recipe storage https://recipe.exchange

When you log into other AT Prototocol powered apps, the data is written to your own PDS. Here's my (Bluesky hosted) account with all the data types: https://atproto-browser.vercel.app/at/bmann.ca

It is also extremely low resource usage. Runs on NodeJS and SQL lite and some web socket connections.

I haven't gotten far in learning Cloudron packaging myself yet, but I'll update this thread if I get something guying.

@rbin saw your ticket on support. I sent you a response there.

Instead of no-op, I recommend atleast just backing up to the same server. Select Filesystem and put location as /var/backups (this is the default).

@nebulon
I think Mastodon's translate button only appears if the poster has set a language for his / hear toot that's different from your language settings.
It could be that the LT API relies on the source and destination language settings that are submitted by Mastodon. That could be an explanation for the errors I received, b/c not everyone sets the language correct. Maybe the DeepL API is more flexible here.

@girish I've disabled backups for now as I try and work out why there is an issue with it. My Linode backs up daily at the moment so worst case I recover the whole system.

I have it up and working with the new Cloudron LibreTranslate app. Running on the same Linode as my Mastodon.

https://forum.cloudron.io/topic/8527/libretranslate-with-cloudron-mastodon-server/1

@shanelord01 you can ignore them. It's just logs from the box code trying to handle well known first before it hits mastodon.

@nebulon Yep - seems to be due to insufficient processes and threads in config.sh

I'm playing with:

#!/bin/bash # Setup scaling related environment variables here - https://docs.joinmastodon.org/admin/scaling/ # Puma export WEB_CONCURRENCY=10 # number of worker processes export MAX_THREADS=20 # the number of threads per process # Streaming API export STREAMING_CLUSTER_NUM=2 # number of worker processes # Sidekiq export SIDEKIQ_THREADS=80 export DB_POOL=100 # must be at least the same as the number of threads

at the moment and it seems to be catching up. 🤞

Yeah, at the moment it's a script that calls the mastodon api to add a git-ops derived block list. Hachyderm has one too. It's a neat idea (philosophical questions not withstanding).

I think the easiest approach would be to download the code locally and have it run against your Mastodon instance and that would allow folks to choose to use it if they want or not, but not bake it into the default app.

But I may have read the instructions wrong!

@shanelord01 You can ignore the recommendation. I think it's a hint for developers to optimize (and maybe sysadmins to report back to upstream project of bug/issues).