They're being targeted for constant DDoS attacks. Could be multiple reasons. But the takeaway should be to spread out evenly through the fediverse. Don't all pile into .world (as I post this on my .world account).
Lemmy Explorer can help give a big picture view of what's out there, both in instances and in communities. Kbin instances are also viewable, you have to select them from the top right menu.
Been on kbin.social for a month and it's been pretty solid. Note it's a different platform than lemmy (designed to read both lemmy and mastodon), but does a great job posting to and handling the content from lemmy platforms. There's growing pains here too, but the uptime has been good.
The big reason for being a target is the size. Why attack smaller ones with less effect? Like making a virus for anything besides Windows. But maybe the long term benefit is the movement of some to other instances, balancing out the loads.
They were getting hit with a lot of DDoS attacks a week or two ago, that may still be going on. Also they took off faster than most instances so a lot of this is just growing pains.
many people have been bailing from the server
That's not a bad thing, the content is shared between instances so you can get it on the other sites too. People going to other sites will hopefully help balance the load and possibly help with the growing pains. I haven't created an account on another instance yet, but I've been visiting other instances when .world goes down.
Massive amounts of organic traffic to too small a server technically is a DDOS, even if unintentional rather than malicious. So, both to varying degrees, probably?
It's DDOS. The admins for World have explicitly said so, and even said exactly how the attacks have been perpetrated by exploiting calls that require a lot of processing time to overload the server.
It's like a fun* little cat and mouse game, you figure out the patterns to block specific traffic, then they adapt and you start again. I even saw some users comment the other day they were being hit with false positives and blocked because the blocking was too aggressive. Fortunately there are companies that specialize in this kind of stuff like cloudflare. But that costs money so it wasn't added until just recently, so it's possible attacks are still getting around that.
Various clues, like what URLs they are requesting, what IPs/regions it's from, if it appears to be real clients based on user agent and ability to execute javascript, and so on.
I've hopped to another server due to the abysmal uptime of late. I'll continue checking in though. I know they're working on adding sysops so I'm sure things will improve.
Use LASIM to download your subbed communities next time lemmy.world is up. Make a new account on another instance. Upload your communities to the new account with LASIM.
Unfortunately, posts/comments don't go - but you can do like me and just leave a trail between your accounts (if you care enough). I just link to and from my old/new accounts in the bio each time I make one. It's a little janky, but it works - and it's better than waiting hours for lemmy.world to be up for 12 minutes before going down again.
There is a status page that tracks the downtime, it's not totally detailed but gives a bit of information about the problem they're dealing with. The last 2 days of logs are particularly bad 14+ hrs of partial downtime.