Hey everyone, so as I’m sure everyone is aware Lemmy.World has been experiencing several outages throughout the last few days.
We have been investigating the root cause of these outages but believe that they are related to our current hosting provider (Hetzner) blocking access from ClouldFlare as (we think) they believe that our CDN is a DDoS’er, and is causing these disconnects to our backend server, problematic for sure.
We’ve opened support tickets with our current provider and are awaiting a response. We have no issue with being as transparent as possible with downtime. Anyone that is curious, can feel free to check out https://status.lemmy.world and https://dash.lemmy.world for up to the minute outage information. We are also looking into other fediverse friendly methods of posting status and outage updates
In the meantime, we are evaluating alternative hosting options and solutions to provide a high level of reliability to you, our users. Really, we want to say thanks to everyone for soldiering through all our technical growing pains.
Cheers
- LW Infra Team
Yeah, pretty big storage requirement due to the way pictrs works. Pictrs is the piece of software Lemmy relies upon to manage image storage, uploads, and most importantly: caches pictures from other instances. This takes up a HUGE amount of storage space, and there’s no official way to clear this up, see these posts I recently made: first one, second one. The solution I resorted to is renting a 1TB storage box from Hetzner for 3 euros per month, pretty sweet deal but I was kinda annoying by it. So the cheapest deal I could find costs me 6 euros per month: 3 for an Alma Linux ARM VPS from Hetzner, and 3 for that storage box. If you’re in for the fun in tinkering (I sure as hell am in), then get ready for a good time. Other than that, if your main line of reasoning is to take burden off of lemmy.world, then I think just go ahead and join another instance. Better yet: join croud funding of another instnace:)
I don’t think this is correct, the images from one instance are hosted on that instance, they don’t get copied over. What’s federated and copied over is text and references to those images (like this
). The only images that are hosted on your own Lemmy instance are the ones that you and your users upload, which sure if you have open sign ups and a bunch of people using your instance uploading images then it will become a problem very quickly, but for personal use it shouldn’t unless you’re uploading a lot of images or even videos.Just use external image hosts for posts and comments, and only use image uploads for your own profile and banner and everything should be fine (if you want extra assurance you can disable uploads in case you forget).
Nah I thought the same but then I manually checked it. In most of the image posts I see, the image URL starts with lemmy.org.il, which made me wonder whether they’re actually downloaded or it’s some kind of whacky proxy. So I downloaded some of these pics and looked for files of identical size and hash digest, and indeed they were on my disk!
It’s not a bad decision to cache pics, because it does make the experience really smooth, and I’m not complaining about it. Mastodon does this as well
Weird because I’ve checked it before and it doesn’t seem to, one notable example is that on the lemmy.zip communities I can’t see any of the images at all because something on my network or with my ISP blocks zip domains. The posts, comments and all their text gets copied over to lemmy.world but the images don’t seem to and thus they don’t display for me due to their domains being blocked.
(Alternate DNS doesn’t seem to bypass whatever they’re doing to block it, I can only connect to .zip domains through Tor or VPN).
I agree it’s definitely a good idea to create a more smooth user experience, would also eliminate the problems that I pointed out with the images not loading due to their domain being blocked.