Eco has been struggling. New players join only to quit cuz of the lag.
However, if the problem is 50% server-side, the other 50% is us, the players.

Would limiting farm sizes be something my fellow eco players would be willing to do? I would hate this personally, I like the 7x7 and it cant be made without at least one expansion so it helps the server, too.
Limiting flush grinder sizes or eliminating them again, just leaving 1-3 so players have somewhere to go for quests? If that is unfair, we need to limit them to 2-4 stacks of 9 layers or something specific like that, or create a /warp hostile.
What about a limit to how many mobs can stack?
What about deciding what a max gold farm size should be and sticking to that? Like 23x23x?x2-4, and each player can only have one. Or would that be too much, too big, too detrimental, still?
Finally, those massive stone gens, are they not affecting our server? Perhaps the server could up the rares and limit the lava sources or something?

Staff, we need some way to access data regarding our crophopper farms. Perhaps a /ch data that shows total revenue and taxes paid since restart that day? Is something like that doable?
We also need way more things added to /warp shop. This would be such an easy and quick thing to do, please consider focusing on that. Put stone, cobble, ores, etc there, too. Anything to get it being used again.

Guys, please comment and add your own suggestions of what we can do from our end to improve things, or what you think the server should do.
It is too late, I think, to say any of these things should be completely banned, but I'm willing to reduce what i have in size if it would improve tps.

 

I’m sure a lot of people with 7x7 farms won’t want the size reduced because 5x5 farms, which is the next viable size down, results in about 50% fewer active chunks. However it needs to be considered that we’re running at half maximum TPS currently. So if reduced farm size and some other changes that Jane is suggesting can get the TPS back up nearer max then the loses will be minimal.

Essentially, a 7x7 farm at 10 TPS will earn about the same as a 5x5 farm at 20 TPS.

Obviously we won’t get to 20 TPS and I don’t know how close we can get if these changes are put in place. I don’t know what other causes of lag there are, other than the obvious ones but it would be interesting to know.

If this can be done and the end result is, say 20% loss in income, perhaps we can have taxes reduced by this much to balance it out?

I feel it’s quite possible to reduce farm sizes and reduce taxes such that our incomes remain roughly the same but with a TPS boost that benefits everyone.

 

I disagree that reducing render distance will fix anything, long term. Almost a year ago, TPS started plunging and our render distance was reduced from 9x9 to 7x7, on the grounds that it would help TPS. And yet, here we are.

My deeply unpopular suggestion would be to nerf moss. And maybe wheat. Take anything lagging the server and nerf it into the ground. Revamp warp shop and add loads of items that can be farmed without causing lag, introducing new revenue streams.

Get spawners added in so that we can get rid of water flush mob farms!

 

When farms were reduced from 9x9 there were very few people with farms that size so it’s not surprising that we saw little change from that. there are a great deal of farms at 7x7 and like all I don’t want to see an income loss but if it can be balanced with reduced taxes then I’m for that.

I disagree with getting rid of wheat farms as it’s the end game crop and I’m not sure if it’s any laggier than Melkin. Note, wheat farms planted in rows are most efficient and have the same number of crops as a melkin farm.

I agree with getting rid of laggy practices like flush farms but that doesn’t need to wait for spawners to come, they can be outlawed now.

I feel that there is a reluctance to outlaw practices and rather things are just nerfed.
Perhaps it’s easier to make certain practices ‘less desirable’ by nerfing prices because there isn’t a great staff presence to keep an eye on laggy or outlawed practices. Sadly those nerfs have a negative effect on everyone rather than just those who bring them about.

 

I fully agree with nerfing or making illegal anything that's lagging out the server. Whatever works, i'm in.

 

It's too late. More limits would have a big detrimental effect for only minor, if even noticeable, benefit. We've been told over and over again we have the best hardware, and regardless of if this is true, I don't think improving hardware will lead to a meaningful performance increase. This leaves only migrating to multi-threaded server software, which is possible, but not without great effort and expertise from a/the development team.

 

The main reason that I don't play eco much is because of the lag. I jump over every so often then end up back on survival pretty quickly because the eco lag is much worse (in my experience).

This is essentially a +1 support for solutions from the perspective of an 'early game' eco player.

 

Flush mob farms have always been limited to 1 or 2 towers, up to 15 layers, since the start of the update last year. Its the reason I took over half of mine down and no longer use the farm. But now some players have been building Island Sized flush farms for mobs, which most likely is what's killing tps. That plus a lot of mapartists use printer and baritone which would also have an affect on server lag. Island sized crophopper farms could be limited, as the croppers detect for growing crops every tick, but then so does tax itself. That impacts the tps. At this point players should refocus from building redstone flush intensive builds, and cropp hopper farms that are left just afk every day and maybe enjoy some other less resource intensive methods of farming money. Cocoa beans, chorus flowers, blaze powder, mushrooms, amongst other farmable items don't use nearly as much server resources to run every day and pay considerably more double if not over what crop hopper farms can pay. It just means actually playing the game rather than sat afking it everyday using up valuable resources.

 

Coincidentally logged into this account and saw this thread while I was checking something, lets take a look at the issue from a technical perspective based on the insight I have from early 2024 when I was still on the team (I think pretty much nothing changed since):

The standard server implementation provided by Mojang is scaling extremely bad and performing even worse. In general all standard server implementations (like Spigot, Paper, etc.) scale vertically which means that you have to improve the hardware resources of a single machine to see an increase in performance. This naturally has a cap especially when you consider that the server implementation is single-threaded which is much easier to implement but also means that the server can't fully utilize all the CPU cores.

This leads to the first question: Could the server hardware be upgraded to an extend where it would make a reasonable difference?

In my opinion yes, at least in theory and without sharing any hardware details since it might be considering confidential info. Something you have to keep in mind is just because there is better hardware it doesn't mean that the hoster (in the case of Skyblock.net) is offering a viable system with said hardware. Lets just say that there are CPUs with much better single core performance (turbo) and better CPU caches.

This leads us to the next question: What exactly is causing the lag/how is the server load distributed?

The average MSPT (milliseconds per tick) for the server based on the most recent report I have access to is 65.93ms, that's 15.93ms over the maximum. The standard server implementation is ticking at 20 ticks a second so in order to maintain 20 TPS you can only spend up to 50ms on the processing of a single tick. Based on those statistics the server is on average running at ~15.16 TPS, this might seem higher than expected but you have to keep in mind that this includes the time after the reboot and times where the server is less busy plus it's an average so kinda meh anyways. The most resource intensive plugin seems to be the Crop Hopper plugin which is (on average) taking ~1.23ms to process per tick. In general plugins only make up for a fraction of the server load, on average for less than 8%. Around 34% is regular entity ticking, around 30% block ticking (including block entities), around 19% player tracking/ticking, the rest is coming from various things adding up. Those numbers are of course not 100% accurate (based on averages etc.) and may change slightly depending on the report but they should still give you a general idea.

Now lets discuss a couple of suggested solutions that have been mentioned:

Q: Couldn't the server simply reduce the simulation distance to improve the performance?
A: Yes, theoretically reducing the simulation distance would reduce the overall server load however it'd also mean that a lot of players would've wasted hours of their time expanding farms to their current size. It's also hard to predict how much impact it'd actually have, you might think that you can simply calculate it (for example by claiming loading 20% less chunks would reduce the load by around 20%) however it's not that simple and in reality I highly doubt that it'd make a big enough difference based on past experiences.

Q: Why doesn't the server switch to a multi-threaded server implementation?
A: Because it's not that easy and because there is no viable option. Some people might bring up the Folia project but it's just a proof of concept with only one contributor. Development is slow and complex, it's absolutely not viable for a production environment. It's too unsafe, unfinished and has a lot of problems. The implementation of a lot of things is also completely missing. On top of that even if Folia was production ready (it's not) most of the 100+ plugins would need to be updated (some more than others) to actually support Folia which is an absurd amount of work that the server can't afford.

So what's going to happen? How can the problem be solved? Realistically: It can't really be solved. Servers are not as profitable anymore as most people think they are and even if the server would invest everything it could into the development to move to an alternative system (which I'll explain shortly) it'd take forever to implement because of the limited funds.

Now lets assume that money wouldn't play a role in this: The probably best and most realistic solution long term would be to split the load onto multiple smaller servers, also a difficult and very complex task. The data of the old island and related player data would have to be extracted out of the vanilla world files and converted to a new world format (similar to Hypixel's slime world format) which stores the entire island as one world in one file that can then be stored in a regular object store (like S3). The island could then be obtained and loaded on-demand on any server (assuming that the necessary system is in place). Communication and things like the tab list would have to be synced between servers to give the illusion of one big server rather than a lot of small servers. A system like this could in theory utilize EC2 spot instances (which cost a fraction of regular EC2 instances and are billed per hour) and could therefor scale extremely well on-demand (assuming that they are powerful enough). This all probably sounds much easier than it is but you get the idea. This should also heavily reduce the workload of updating the existing plugins.

 

matt has said that entities are the main cause of lag

 

How much of that is scheduled ticks, like those caused by fluids and redstone?

I think everytime someone has to work around the hopper limits (and speed). With alternate methods, it defeats the porpose of this imposed limits, and at the same time takes a route that is probably a bigger resource hog.

Examples:
minecart hoppers,
droppers to move items.
using MORE hoppers.
etc.

 

don't ask me, i personally think that minecart hoppers should be more limited or removed all together

 

They are required for the only sorting system that works reliably on the server.

 

if there was a sorting system i could use i'd use it. people have gone without sorting systems for long enough and a way to fix this would be adding an alternative or easier way to do this.

 

People who don’t know how to build them have gone without them.

 

and if most people don't know how to build, then we can just add an item which helps with that issue, everyone gets their sorting systems and there's less lag.

 

Yeah, would be nice but I have no idea when the last suggestion that required coding was responded to let alone implemented so I'll just keep using the system I have.

 

Sorry for the necroposting, but would you guys ever reconsider the last option? I've noticed a lot of new players don't stay long-term due to server lag. Would it be worth investing the money now towards that, and then making it back later? I'm sure you guys have exhausted an extensive list of options, but the recent constrictions with little effect on tps have me and a lot of people worried about the server and the lasting impacts on the community. Several people have quit, and even more are considering quitting due to the recent updates (alongside general server negligence) within the past year and a half. Speaking from an owner and business standpoint, those are your supporters and customers you're losing out on. I am angry, but I'm also very concerned.

 

thank you for the response :] very happy to hear you guys are working on it