this post was submitted on 05 Jun 2023
156 points (100.0% liked)

Lemmy

496 readers
1 users here now

Everything about Lemmy; bugs, gripes, praises, and advocacy.

For discussion about the lemmy.ml instance, go to !meta@lemmy.ml.

founded 4 years ago
MODERATORS
 

With forewarning about a huge influx of users, you know Lemmy.ml will go down. Even if people go to https://join-lemmy.org/instances and disperse among the great instances there, the servers will go down.

Ruqqus had this issue too. Every time there was a mass exodus from Reddit, Ruqqus would go down, and hardly reap the rewards.

Even if it's not sustainable, just for one month, I'd like to see Lemmy.ml drastically boost their server power. If we can raise money as a community, what kind of server could we get for 100$? 500$? 1,000$?

you are viewing a single comment's thread
view the rest of the comments
[–] poVoq@slrpnk.net 8 points 1 year ago (1 children)

The advantage kbin has is that it is build on a pretty well known and tested php Symphony stack. In theory Lemmy is faster due to being built in Rust, but it is much more home-grown and not as optimized yet.

That said, kbin is also still a pretty new project that hasn't seen much actual load, so likely some dragons linger in its codebase as well.

[–] sam_uk@slrpnk.net 2 points 1 year ago (1 children)

I think it's probably undesirable to end up with big instances. I think the best situation might be one instance that's designed to scale. This could be lemmy.ml or another one. It can absorb these waves of new users.

However it's also designed to expire accounts after six months.

After three months it sends users a email explaining it's time to choose a server, it nags them to do so for a further three months. After that their ability to post is removed. They remain able to migrate their account to a new server.

After 12 months of not logging in the account is purged.

[–] andrew@radiation.party 2 points 1 year ago (1 children)

Thought on this a bit more, and I’m thinking encouraging users not to silo (and make it easy to discover instances and new communities) will probably be the best bet for scaling the network long-term.

“Smart” rate limiting of new accounts and activity per-instance might help with this organically. If a user is told that the instance they’re posting on is too active to create a new post, or too active to accept new users, and then is given alternatives, they might not outright leave.

[–] sam_uk@slrpnk.net 1 points 1 year ago (1 children)

That might work, is there some third party email app that could capture their email and let them know when registrations are open again? I know of some corporate/not privacy respecting ones such as https://kickofflabs.com/campaign-types/waitlist/ but presumably there's a way to do that with some on-site tools?

[–] andrew@radiation.party 2 points 1 year ago (1 children)

If the instance in question has email support, I don't see why the instance couldn't notify them directly - but I think providing alternative instances first (with the option to get notified if this instance opens up) would be more reasonable

[–] sam_uk@slrpnk.net 2 points 1 year ago

The idea would be to retain the ability to collect email addresses, beyond the point that the main app can't keep up. So you'd want something lightweight just for capturing the emails.