actually I did delete the server (after creating a snapshot of it) a week or so ago. But this morning I wanted to check lemmyfly.org, couldn’t load the page. Checking my Hetzner dashboard I noticed CPU was spiked at 200%?! It did drop again though, but apparently had last for 2-3 minutes. But prometheus was down, so no graphs apart from the hetzner ones. I doesn’t relate to network traffic spikes, so I don’t know what caused it. I’ve started the prometheus server again (that snapshot was really useful :) ) and will leave it on for a couple of months now.

current system consumption:

I might need to get an extra volume for storage, Lemmy is starting to eat up the root filesystem… Does anyone know how I re-configure Lemmy to look at a different volume for storage ?

http_response metrics are also up, using https://github.com/martin-helmich/prometheus-nginxlog-exporter

would be nice to see the numbers go up :)

Let us know your findings when you did!

i’m now looking into logging more specific http request data. Maybe https://github.com/martin-helmich/prometheus-nginxlog-exporter?

I use namecheap’s email service privateemail since a couple of years. Google workspace was great, but don’t need the other features anymore and this is fifth of the price https://www.namecheap.com/hosting/email/

Looking into the database, it contains many thousands of posts. I’m assuming this is stored in the local db for serving it to instance members. So when you open a post from instance B on instance A, A fetches post-data from B, stores it in A database, then serve the content from db A to the browser

every instance is sharing in the traffic to browse the fediverse. Not one service is responsible for serving content, you (the instance admin) are only serving for your members.

The downside of this is there is a huge amount of replicated data stored everywhere. Content of popular communities will be scraped by and stored on many many servers, filling up servers and increasing storage and bandwith bills for all those servers

no, a random image I just now check was 3.3mb even. But you’re right, the image source does point to the origins server location… Then why are all those images stored on the server ?

When I look on my server in the volumes/pictrs folder, there are a LOT of images that I recognize from my feed. They are not from posts posted to my server but belong to posts of communities that I have subscribed to

Thats the way the decentralized fediverse works (I think!) let me know if I’m wrong

only pictures and posts that the instance-member are subscribed to, and from the moment they interacted with it. But yes, I’m curious to see where this goes.

I noticed it when I was looking in the database and disk on my instance, where there are already thousands of posts and more then a thousand images - while my own instance only has a couple of posts and images

lots of traffic on many small files eventually uses more disk space and bandwidth. Depends on the growth of the instance

yes. running a server costs money. I have lemmyfly.org running just 2 days now on digitalocean using a cheap 1GB ram, 25GB diskspace and 1000GB bandwidth for $7,- a month. Storing images on the server will eventually take up all space and bandwidth meaning you have to upscale -> pay more.

Using a different location for storing images and/ or videos is best to offload the instance !

yes, they are hosted on the server that is running the Lemmy instance. But every other instance that is linked to that instance will ‘scrape’ all posts incl images and store them on they’re own server. So posts and images are served from the instance you yourself are a member of

I do, and looking forward to contribute. I’m not a UX designer though…

it definitely needs some UX attention