Just in case anyone is using their account here to post to off-instance communities: those posts and comments seem to have a very high failure rate. There is a lot of activity and accounts on this instance. This is to raise awareness, not to pull people away or break up the lemmyverse. Quite the opposite really: there is a technical problem on this instance that might be preventing the lemmyverse from functioning as it should.
Lemmy.world and some others have held back on the 0.18.0 back end update because the admin did not want to forego this Captcha integration, which broke in 0.18.0. Version 0.18.1 is expected to release soon, and address the regression with Captcha.
The new version also reduces its reliance on websockete, which should address a number of other quality-of-life problems on this instance, such as the random post bug (and hopefully the 404s/JSON errors I’ve been getting all afternoon).
Hopefully just a few more days and we’ll be back to rights.
The new version also reduces its reliance on websockete
From my understaind it’s not a reduction, but a complete repalcement.
You’re probably right. I’m not a programmer and am not familiar with the code base so I was avoiding speaking in absolutes.
I don’t think it’s a version issue. It could be, but testing I’ve done says otherwise. There’s plenty of 0.17.4 instances that, despite other bugs, have no problem sending and receiving updates from other instances. I hope you’re right. I don’t run the instance so testing reveals mostly speculation.
There’s plenty of 0.17.4 instances that, despite other bugs, have no problem sending and receiving updates from other instances.
It’s probably because of the absolute scale of dot world. This is what happens when you try to centralize ActivityPub, especially an implementation like Lemmy which did not previously have any of the scaling challenges the likes of Mastodon had before.
Everyone should’ve told people to pick different instances (there are STILL new guides written where step 1 is to “just register on dot world” [or shitjustworks, which isn’t any better, really]) and admin should’ve locked registrations after 10-15k, maybe 20k users MAX, yet y’all got too greedy. Good fucking luck dealing with the aftermath.
I should probably go to bed, I’m getting grumpy.
In my opinion, we need to somehow solve the community centralization issue first. MultiCommunities, or some way to aggregate the dozens of large-ish groups like “news, technology, etc” and be able to subscribe to all of them in one fell swoop would allow people to spread out to other instances much more reliably.
I’ve brought this up as a suggestion elsewhere. People seem annoyed at the lemmy vs kbin idea of “communities vs magazines”. Maybe everything is changed to “communities” and “magazines” are officially adopted as community-maintained mega-lists of common communities.
An example. There’s a bunch of car manufacturers. Sure, maybe I could just select the “Honda” community on every instance I can find, or instead I subscribe to the magazine called “Honda” which auto-subscribes me to every single Honda community in the list… or even the magazine called “Cars” which would include all manufacturers and cars communities.
Then there could be a Magazine view for that Magazine which would allow all posts from those communities to be aggregated in one place.
Just spitballing ideas.
It’s a known issue:
https://github.com/LemmyNet/lemmy/issues/3101
Seems to be growing pains from Lemmy having to scale up in a very short amount of time so hopefully they are able to find a workaround soon.
It’s not happening everywhere and all the time but it happens enough for you to think a post or comment has no engagement when it does just not on your instance.
It’s a known issue, but if you look at the issues, the reason for it is “could be lots of things.” Relying on a FOSS project to fix the problems of the community is kind of naive. If the admins don’t step up then I’ve no sympathy for them when the whole thing crumbles.
Why did this get downvoted? Something being an “issue” on GitHub is about as meaningless as something can be. Unengaged developers are usually the result of unengaged users. I’m saying IF your admin (not that they did,) submitted an issue and then is just sitting back and waiting, that’s shitty.
That isn’t at all what is happening, the world admin team has been very engaged with the project on GitHub with a ton of back and forth and various code pushes to fix the captcha pullback that’s in 0.18. The issue you are seeing is known and the belief is it should get better with the update to 0.18.1 that the devs have said is coming in the next few days. It seems to be partly due to the size of the world instance and the problems with web sockets. You’re likely being downvoted for appearing to be authoritative about this and blaming the world admin team (it’s more than a single person) when they know this is a symptom and have a plan they think will fix it that a little more digging into the history of why they’ve chosen to forgo the 0.18.0 update would have answered your questions.
What do you think is causing the failures? How are you seeing the fact that it’s failing?
Like you mentioned afterward. Comments and posts just plain failing to land on any other instance. Also I run an instance for testing and can see incoming connections. lemmy.world fails at a protocol level, not at the application level. It’s a, IMHO, bandwidth issue. Hopefully the admin is aware and wants to fix it. I’d say he has a responsibility, but he doesn’t. lol.
Also keep in mind that defederating (blocked instances) will prevent posts and comments from syncing between instances as well.
You can see blocked instances on lemmy.world here: https://lemmy.world/instances
