With the sluggishness of the site these last few days you can sometimes get slowness or infinitely loading icon when hitting submit. If you aren’t sure if your post submitted, refresh the original page in a different tab before hitting submit again to see if it went through. I see double and triple posts in almost every comment section.
Yes! The Lemmy devs should make it so that you can’t hit the submit button 10 times in a row. People hit submit and nothing happens so they hit it again and again and again. Now there’s a bunch of identical posts.
I remember the old days where sites would have a message saying “only hit submit once”. That’s gone now since sites prevent users from hitting submit more than once. I don’t think Lemmy is doing that though!
It’s problematic because often you’ll get an error that says it failed to post the comment. In which case it’s fair to try again, but it actually posted anyways.
Just get off .World and onto another instance and it’s smooth sailing with mobile apps working great too.
programming.dev is getting to be pretty damn slow too…
Comments post in 5-30 seconds, and up votes can take 30+ seconds to post. Searches are pointless, and loading comments can be a repetitive painful process of reloading till it works (504 gateway timeout)
This is especially bad on really popular instances like lemmy.world. It seems the server can’t keep up with all the requests or something along those lines.
Is there a way to load balance these lemmy instances across multiple backend servers?
Yes, and I think they are.
Backend sizes likely need to be larger and more abundant. Depending on hosting provider this could be mean lighting your wallet on fire every month till you either get enough donations to make up for it, or you give up and shut it down.
This problem will only get worse as Lemmy increases in popularity from “extremely niche, extremely low volume”/trivial volume (~50-100k users) to anything other than that (1mill - 100mill users or further). And gets more difficult since economies of scale may be difficult to apply here due to the hosting model of Lemmy, meaning cost/user is likely to stay quite high when normally it would be drastically reduced the more users you have.
I’m gonna guess individual instances are going to be forced to break down into instances that host only a couple communities each as time goes on. This would be the pragmatic choice since it’s very likely that individual communities can grow large enough to tank Lemmy instance performance for all others on that instance. Although that presents non-trivial UX concerns.
Though, again, many smaller scale instances means infrastructure costs may stay prohibitively high as a result of over provisioning or a lack of value from aggressive caching, read replicas…etc 🤔
Edit: I wonder if an instance can host individual communities on their own hardware. Or if there is some abstraction that lets users browse potentially hundreds/thousands of instances easily as if they where all communities?
