Why do you think that google, Microsoft and Amazon never got into the infrastructure business before?

Amazon was in the infrastructure business well before containers were the “big thing”.

Podman rootless, using quadlets for systemd services. :D

Check out the following link - I am pretty sure its what I used to get it all working.

https://3os.org/infrastructure/proxmox/gpu-passthrough/igpu-passthrough-to-vm/

Hey, sorry for the late reply. I am running rootless using a dedicated user, so I use systemctl --user to control the container. From what I understand, when running rootless the root user inside the container correlates to the outside user (which is running the container), in terms of permissions. The external directories I bind mount into the container as externally owned by my dedicated user, so that the root user inside the container owns them (inside the container).

Are you doing rootless or rootfull podman? I am doing rootless and I have the following in my radarr container - PUID=0 PGID=0

I am using Calibre-Web mostly - but I have run into issues with thumbnail generation after my collection hit around 500000 books. I am just over 600000 now, but a large swathe don’t have thumbnails unless I do a manual metadata search. I should probably look for an alternative, but at this point I CBF.

Yeah it looks pretty slick but not so much slicker than Frigate that I will pay to be in the beta. :)

Personally I would lean towards finding out why its borking with SELinux and fixing that. It really shouldn’t be too hard. As others have mentioned it may be as simple as how you are mounting volumes into your containers - or it could be changing the SELinux context type for some files.

Yeah this was it. Disabled rocket and it now works fine.

Yes I bet this is it. I’ll disable and test. Thanks for the heads up!

Thanks for these. It’s good to see someone else building them!

I have the same problem using your 0.18.3 lemmy-ui image as I do when I build it - it just doesn’t seem to work on my instance. None of the feed loads up and the selector buttons don’t either.

Is it working for you?

I’m doing that. 4 core arm instance with 24GB ram. It’s on a paid account but using free tier.

From the docs / troubleshooting:

“Also ensure that the time is accurately set on your server. Activities are signed with a timestamp, and will be discarded if it is off by more than 10 seconds.”

I am not 100% surprised they refuse to do it for new accounts. If you have an account that has been with them for a while, they most likely would open it.

Problem with SES is that you start sandboxed and can only deliver to specific email addresses - which obviously won’t work here.

I didn’t bother, as I was just testing. But you are right, port 25 outbound is blocked by default. They have a defined process for you to ask for it to be unblocked and you have to tell them what you are using it for and how you are preventing spam from being sent. In this case it might be enough to say that you aren’t allowing port 25 inbound, so it can’t be used as an open relay.

Looks really good. I did it pretty much the same way, myself - but if I were looking to start again, I would definitely use this.

Edit: Ran it on a fresh AWS Ubuntu instance and it worked perfectly fine.

Sorry… I am not understanding fully, I think. So you want to see if posts on your self-hosted instance will propagate to other instances? In this case, only if someone on the other instance has searched for your community.

It’s not distributed architecture as you normally think it - it’s a decentralised federation. It’s an important distinction from your typical distributed architecture app.

Thumbnails are stored locally, I believe.

A bunch (47 containers at present)… Won’t list them here as its kind of redundant with what a lot of other people are running. My latest is Lemmy (lemmy.nine-hells.net).