Has anyone tried to deploy the docker imagine with AWS ECS? If yes, could you please share how you have done it?
I’ve tried searching this forum for instructions but with no avail. Things that I’m trying to figure out at the moment are how to create a space (should it be S3 or can it be a folder?) and how to expose the space to the internet.
I doubt anybody has attempted this, but let’s see.
A few words of warning:
Although there is indeed support for using S3 as storage for your space files there’s a few reasons not to use it:
- It’s not been very well tested and I’m not sure anybody is actually using it “in production”. It’s a thing I built when I was still actively pursuing a way to turn SilverBullet into a SaaS-style product. I don’t have such plans now. This also means it’s on my target list to ultimately be removed, if it becomes a maintenance burden.
- If you use this, you cannot use any plugs that rely on running shell commands (such as the Git plug, or @Maarrk 's git grep plug) because those will assume physical files on disk.
- It still requires a database to keep meta data about files, and the only viable option is to use Deno’s KV, which uses a SQlite file on disk. This mean you will need a way to have a persistent volume attached to your container anyway. And if you need that, you may as well use it to keep your space folder as well.
As a logical consequence, SilverBullet is not horizontally scalable in its current form at all. That is: you can only run it in one container at a time attached to one persistent volume.
I’m no ECS expert, but this would mean that you need to configure it somehow:
- To only ever run in a single container at a time
- That container should be connected to a single persistent volume that keeps your space (both the DB file as well as space files).
Now it’s for you to decide if this is a viable setup for you still 
Just commenting to mention about s3fs-fuse/s3fs-fuse. It is also a hacky solution, but thought I would bring it up.