- cross-posted to:
- tech@pawb.social
- cross-posted to:
- tech@pawb.social
Some of the planned blackouts will be temporary, others plan to shut their subreddits down indefinitely in protest.
Some of the planned blackouts will be temporary, others plan to shut their subreddits down indefinitely in protest.
Problem: millions of redditors currently use third-party Reddit apps. Abruptly sending millions of people to the Lemmy instance you just deployed is a sure-fire way to break it, and maybe bring down the whole federated network of Lemmy instances. Lemmy currently has issues scaling above a few hundred users, as Beehaw has recently discovered, let alone millions.
Problem: Lemmy is a completely different protocol, and there’s less than a month left before all third-party Reddit apps become useless and everyone uninstalls them. That’s an exceedingly tight timetable and an exceedingly unforgiving deadline.
That said, it’s now or never; death or glory. We’re not going to get another chance to bring over that many people to Lemmy all at once.
@argv_minus_one Is that scaling problem a software issue, or a hosting issue? There are other Fediverse platforms like Akkoma that use Elixir, so maybe they’d fair better? Could also pick several federated instances to distribute users to.
Lemmy is written in async Rust. The language isn’t going to create a scaling problem. Well-written async Rust applications have handled vastly heavier workloads than Lemmy without a hitch.
There are, however, some serious performance bottlenecks that need to be dealt with, and it remains to be seen whether any more bottlenecks remain undiscovered in either the protocol or the implementation. To be honest, as someone working on a Rust+Postgres application myself, this is the sort of thing that keeps me up at night.
Hosting can of course be an issue as well. I’m under the impression that Beehaw had to go up several tiers in its hosting plan in the last few days in response to the surge in demand. I assume this was done to work around the aforementioned bottlenecks by simply throwing more hardware at the problem, but I don’t know.
@argv_minus_one I see. The more I look into it, the more I think Lemmy should still be considered beta software like kbin, TBH. Some important features are still missing and the optimization is lacking.
I do consider Lemmy to be beta software.
But it’s currently the best option.
@ericjmorey You ever look at kbin? It already has the follow user functionality implemented so you can get content from the rest of the Fediverse.
My biggest concern with kbin is its in alpha, very little time in use to smooth things out, there’s more likely to be major changes to its functionality, possibly breaking changes. I’m olso not too interested in following individuals outside of my Mastodon account.
@ericjmorey That sounds like the situation with Lemmy, but with less features.
I’m not sure I’m understanding. Lemmy is different in the way I described. What features (other than following individuals) are you talking about?
Yep. It works 🥳 and has all the basic features you’d expect of a Reddit replacement 🥳 but will no doubt have the same growing pains as early Reddit did.
@argv_minus_one I still remember when Reddit was 503ing on the regular.
Wasn’t that just last week?
I wasn’t able to load threads or comments in RiF not a few hours ago.
I still get that stupid “You broke Reddit!” screen all the time.