Hey friends, it's hard to write this, but it's time to retire botsin.space. I wrote a post about it here: https://muffinlabs.com/posts/2024/10/29/10-29-rip-botsin-space/
-
@GossiTheDog @muffinista We will need more details about this, please open a GitHub issue and we will have a look. But we have not seen anything like that on our side, so I am really curious about what is happening here.
-
gellenburg :heart_pride: :verified_gay:replied to Sean Tilley last edited by
@[email protected] @[email protected]
File storage and bandwidth. These expenses will also only get more expensive over time. File bandwidth is the #1 charge on the monthly server bill right now. I live in fear of an AI scraper figuring out how to scrape all of these files and bankrupting me overnight.
This is fucking infuriating!
#Mastodon is the only fucking platform on the #fediverse that forces #instance #admins by default to fucking cache every single goddamn media file that crosses paths with their server.
File storage and bandwidth costs are the single LARGEST expense for ANY Mastodon instance.
I help run two of the largest NSFW instances on the Fediverse and our file storage and bandwidth costs are so low we haven't even asked our users to help pay for them (yet).
Why? Because we're only responsible for hosting and serving content our users upload, not every other user on the fediverse.
Jesus. Fuck Mastodon for this stupid and asinine behavior. -
@renchap Would be interesting! Not sure whether muffinista has the energy for this now. https://muffinlabs.com/posts/2022/09/10/how-i-maintain-botsin-space/ mentions that the conversations and status tables are especially massive, and I imagine the social graph too given the number of followers. Maybe the new version's follow recommendations are expensive in such a situation? Does the notification grouping query the DB to decide those groups?
-
Sean Tilleyreplied to gellenburg :heart_pride: :verified_gay: last edited by
@gellenburg Yeah, when we ran a test instance a while back, we found ourselves dealing with a constant overflow in our Object Storage buckets. Just absolute fucking garbage, you’re apparently supposed to routinely run a cleanup job every two weeks, and basically mirror whatever comes across your timeline.
I understand that there are trade-offs, that this is supposed to help the timeline or media loading or something, but it’s a bad default and fucking sucks. To my knowledge, no other Fediverse platform does S3 storage this way.
-
@muffinista that's quite sad news, but I understand it.
Is your fork of Mastodon open-source? (it should be but y'know)
I'm thinking maybe the next-generation botsin.space could be running something more specialised for bots?
e.g., something written on @fedify instead of mastodon
-
@thisismissem my changes are really minor but I will publish them. I do think a platform that is more bot-oriented is a good idea
-
@muffinista cool, I look forward to seeing them & better understanding needs.
-
:PUA: Shlee fucked around andreplied to colin mitchell last edited by [email protected]
@muffinista @colin please find another admin to take over instead of burning it to the ground.
-
@muffinista I am following somme of those bots, so thank you for running the server for as long as you have!
How about a shout-out to have everybody set their bots to autodelete older posts? That might save you some expenses on server space for the remaining time.
-
@nemobis I doubt that follow notifications or notification grouping are an issue here if there are not a lot of real users using the web UI. I can not guess what the cause may be without more informations, hence the need for a GitHub issue and more data.
