I’ve been playing with the self-hosted instance of Speckle managed through Coolify in a Hetzner VPS and found it was surprisingly easy to deploy and manage. Just had to adjust the Docker-Compose and was pretty much ready to go. Anyone else has tried before? Would love to discuss the technicals of it and understand the limits. This feels very game changing, but suspisously easy hahaha
Coolify is super cool ! I allows you to host many applications on the same server for free, making the management process so simple.
I am on the process to replace multiple subscription with an open source hosted version.
In the past, I self-hosted discourse or ghost and ended in some dev ops problems and finally choosed the official hosted version (for ghost at least).
But now with coolify it is clear it will change…
You can even us it to host your own apps, and it handles the deployement, dockerisation…
It also works with multiple server.
You can have some db as a service as well, all in open source.
With an automatic backup to Amazon S3.
I made a try with speckle passing the github adress of Speckle server, it failed, but maybe I choosed a wrong option…
If someone could try with more knowledge of speckle than me ?
To understand coolify you need a fresh server with at least 8Gm RAM and 2 CPU…
I choosed one new in hostinger for my part, and I already have 4-5 services on it !
Note: We’re generally very keen as a company to serve servers for you. That aside, personally, I don’t have any experience with coolify and we haven’t tested our server against it - though the project does look very cool!
Eyeballing it, to run the full server with its associated dependencies and services (redis, pg, etc preview, file importer) - you might need a beefier VM.
I still haven’t gotten closer to actually playing with coolify unforch, so no exact pointers from me either.
I’m not sure I fully understand what it means pointing it to the github address of the server.
Does coolify have an option to just try to bootstrap an app deployment from source?
If yes, we’ve def not optimized for that and given we’re doing some (relatively) complex stuff with our monorepo and dependencies, I would not expect that to work with our current state.
It is a cool(ify) idea though!
ps:
Dim is right, you’d prob also need more ram to run the full server stack next to your already running services.