When you choose a Lemmy home instance, you’re trusting somebody else’s infrastructure with your account, posts, comments, and uploaded media.
If that instance disappears without a usable backup or migration path, years of your contributions can disappear with it.
Yet when I look at most instances, I have absolutely no idea what I’m trusting.
Is this running on:
- A $5 VPS?
- A dedicated server with redundant storage?
- Someone’s mini PC under their desk?
- AWS with costs that explode exponentially as traffic increases and hapless admin not realizing that just shutting it down as they never had a plan?
- A database with proper backups?
- A server with no backups at all?
I recently had an instance tell me my posting activity alone was becoming too expensive for them. Their solution was - just banning me and removing all my data. no im not kidding - thx https://lemmy.radio/.
That immediately makes me wonder: what are you running? (apparently it was aws micro or something and person clearly had no idea what they were doing)
If a few hundred posts can create a meaningful financial problem, is the instance massively undersized? Are you paying enterprise hyperscaler prices for workloads that could run cheaply on dedicated hardware? Is storage the problem? Bandwidth? Pictrs? Database load? Do you even have a plan my friends?
I don’t know, because users are given almost zero infrastructure information.
I’m not asking admins to publish IP addresses, network diagrams, credentials, or anything security-sensitive.
Just put some basics in /about:
- Hosting: OVH / Hetzner / AWS / home-hosted / etc.
- Compute: 8 cores / 32 GB RAM
- Storage: 2 TB NVMe
- Backups: Daily, retained 30 days
- Media backups: Yes / No
- Off-site backups: Yes / No
- Approximate capacity: Small / Medium / Large
- Funding: Self-funded / donations / nonprofit / etc.
That tells me infinitely more about whether I’m comfortable making an instance my home.
- Have admin(s) ever hosted any production software before? First time?
- are there even backups?
- Is there even any plans for scalability? What if your traffic goes 4x? what about 20x? do you have a plan beyond banning me and removing everythig I ever posted?
And honestly, backups matter to same tune as CPU model.
Federation gives us the freedom to choose our instance. Great.
But you can’t make an informed choice when you have no idea whether your digital home is a properly backed-up server or a $5 VPS from a racknerd black friday sale held together with Docker Compose and prayers.
Is that the reason why all lemmy instances dont let you upload more than 5Mb media? probably.
If you cant afford hosting costs - holy fuck, just stop fucking saying “everyone welcome” if you then going to ban those same “welcomed” people after they post 10 image memes (thx https://sopuli.xyz/ and https://lemmy.zip/ and https://lemmy.radio/ and https://lemmy.ml/). no no - 👏you 👏 cant 👏 afford 👏 non-text 👏 users 👏, - instead you falsely accuse people of spamming and banning their accounts. No wonder lemmy is at 35k MAU. Hardware poor admins - you don’t even know what spamming is, you shouldn’t have kids users if you cant afford them, dont invite them in or just advertise your instances as text-only.
Hardware poor admins - stop inviting people. you have no capacity and only actively harming lemmy adoption.
There should be a list of such instances that should be inviting people at all.
Instance infrastructure shouldn’t be a mystery.
If you’re asking users to build years of history on your server, give them enough information to understand what they’re building it on. You likely cant afford non-text users anyways.
This is a problem with Lemmy on the admin side. I set up retention, and any post here is automatically deleted if it’s more than a year old and nobody has interacted with it. But I had to write a lot of custom scripting to handle that, and it still screws up: I was somehow counting profile photos as posts, deciding they got no interaction, then deleting everyone’s profile photos. I fixed it, of course, but that isn’t something admins should have to handle from scratch, by hand, on every server. No instance owner can afford to store everything, forever, for free, for users. Especially in this market where memory and hard drives are soon going to cost as much as my house. This instance is funded entirely by me, at about $200 per month, located in a Canadian data center. But $200 is a hard cap; I need to find ways to ensure we stay within that monthly amount, or try and raise money from a collection of my fellow blind users, who likely have nothing to spare. This isn’t a beg post, just trying to give you insight into the realities of instance hosting. Right now Lemmy expects that all admins are okay with media and database storage that expands for ever and ever, infinitely. Most of us are not. Meaning that if we’re technical enough, we can write our own custom solutions to handle it. Or if we’re not, our only recourse is banning users to force Lemmy to delete whatever’s filling the disc.
I’m a well federated instance, so I guess I could make a contract to sell all the data I get to an AI provider, and that would allow me to afford to store everything forever. But I’m not willing to do that, and I suspect if I tried every other instance would defederate us anyway.
Since you already put in the work do develop the script, have you considered upstreaming it and/or releasing publicly? It’d help other server admins
No. I currently support and maintain multiple accessibility repos. That’s my area of expertise, so the place most of my opensource time should go. Any server admin could release the system they created for removing old Lemmy posts. Far fewer can write accessibility tools and addons. So I spend my donated time where I can get the most impact for what I put in.
Thanks for all you do. I appreciate your efforts!