Chests, Lag, and a Third Thing I Can't Think Of...

[PosterAnonymous Profile]

This is mostly for those who spent time troubleshooting the Flea Market, but anyone with expertise in the area is free to respond!

I have the beginnings of an idea for a central bank/repository/public storage area and I'm trying to get an idea of the lag implications this would represent.

The main problem we're currently having to deal with is that, while convenient, the Flea Market had to be closed due to the sheer number of chests in that area. As a result, players are having to keep private caches of chests all over the map and travel great distances (in some cases) to restock/etc. What I would like to do is find a relatively unpopulated area close to spawn that could be used as a central storage point for anyone so inclined. This would negate the lag of having copious amounts of storage in one area that is heavily travelled, but it would also negate the need for each player to devote a significant amount of home construction to storage. Think "Gringotts", if you're familiar with Harry Potter.

Yes, I realize the lag in whichever area this would be built in would be significant. That is why I'm wanting to do this in a relatively unoccupied/unused area. SignShop signs linked to the repository chests could then be placed in stores (as they are currently linked to private storage) and I believe a Donate sign could be used to replenish supplies remotely as well.

From my research, it seems each chunk is 16x16x256. In addition, it appears that up to 10 chunks in all directions load per player. What I cannot find is the number of chests per 320x320x256 area it takes to create lag problems. I've asked a few players and no one seems willing to respond. Does anyone have this information? Also, do you all think this is a good idea as long as I avoid populated areas and have a reasonable travel distance from spawn?

In other words, if I build it will you use it?

EDIT: My current plan is to make it free to use, but it will require reservations/a check-out system with the understanding that if you are deemed inactive (have not logged into the server in 3 months and have not notified the forum/myself that you will be away for an extended period of time) your chests will be removed and the items dispensed in some form or fashion (haven't got those details worked out yet). I would attempt to contact you via /mail and PM's on the forum at least twice before taking such action.

[batstarwars1 Profile]

Would there be any cost in using this storage facility?

[PosterAnonymous Profile]

Heh I knew that was going to be *THE* question! I updated the first post, but to summarize: No.

EDIT: Also, I do have the area I want to use selected. I am trying to negotiate with a player on the forum and do not wish to disclose where for obvious reasons at this time.

[batstarwars1 Profile]

Ok, this sounds like an AWESOME idea though. I hate going to the boondocks every time I want to stock my chests up. I live pretty far away from spawn with BO$$, Blacky, and Greg. When would this project get started? If it does, I would be willing to build/gather materials to help make the storage facility.

[PosterAnonymous Profile]

At least after the Aquadome completion. As I said, I have a few things planned immediately after that build is complete and there is someone currently on the site I wish to use. If I get the OK from that player or they are deemed AWOL, I will need to speak to Timber/Skyrail and Robrotheram/D.O.T. about doing this as the spot I have selected would lag a section of D.O.T. and the Skyrail.

I really hope I hear from this player because they have done a lot in the area terraforming-wise and I would hate to just outright take it from them.

[PosterAnonymous Profile]

I have been informed that I should hold off on this idea (not a problem considering this was just in the pre-planning stages). I will wait a bit and see if anything further develops.

[Curley]

What was the final verdict regarding the lag felt by players in the flea market? I know it was discussed to death many times but was it ever fully determined if it was the cheats linked to sign shop, the signs linked to sign shop, or even sign shop at all? There was a time we thought it was wonky lighting issues.

My experiences were, my main comp with 64bit java and 2gbs of memory allocated to java never had any slowdown within the flea market, but my crappy house media center laptop with 32bit java and barely enough memory to run minecraft would practically lock up when walking near there. I only state the stats based on assumption, not testing.

What I'm getting at here is, and don't count this as fact that were doing some drastic change, but, if sign shop is for sure the issue, there are alternatives. We can scrap it and roll out something that doesn't drag down the server while performing its one single function. Sure it'll mean a lot of work to re link shops and refactor how we sell but, I dunno, thought it was worth the discussion here. Before you build a big ol thing.

[Timberwolf1777]

Disclaimer: The conclusions drawn up in this reply are based entirely on in game experience and discussions with other players. Any/all of it may be false.

To the best of my understanding, the signs for signshop are the main issue, especially with older signs that have gone through several updates. I dunno how signshop is programmed but I can only assume that the linking information is stored in some sort of flat file that the server has to use to search for the correct link for each sign. This said, with a million signs, & older ones at that (that may or may not have a different way of being handled cause of signshop plugin changes with the updates) in a few chunks, I cant even begin to imagine the overhead all of that customized info (not to mention each sign being an individual string facing different directions & rendered in a 3D volume) would have on any client PC. It may simply be that a zillion signs is the main issue ... and then add the signshop plug-in's needs to the mix ....

Thoughts?


PS Here is a link for a forum post regarding sign lag:
http://www.minecraftforum.net/topic/843 ... me-people/

[PosterAnonymous Profile]

From a few of the private conversations I have had with some of the players on the server, chests were the main problem in the Flea Market area. A lot of the lag, as Curley suggested, depends upon the client machine. Nice, new PCs handle an area with 2000 chests a lot better than Mom and Dad's 5 year old hand-me-down PC. The problem is we aren't all as rich as your Rich Uncle Poster and therefore not all of our PCs can handle it nor do we all have the ability to upgrade. I'm sure signs come into play as well but the big issue seems to be chests, hence the central repository idea.

Like I said, this part of my idea is being put on hold anyway - I'm talking to peeps.

EDIT: Yes, this means I had other parts to this plan which *ARE* still in-progress behind the scenes.

[Curley]

You are right timber in your assumption regarding sign shop, it's one of the only plugins we have that doesn't use our SQL database, and sadly also one of our most used plugins.

The flat file is horrendous. Outside of all the lag discussion, I'm constantly having to coddle the damn thing to make sure it doesn't crap the sheets. Reference to that one time we lost days of signs shops and I had to dig through manual backups and hand copy paste bit by bit of data to find the corrupted values. I'm always looking for an excuse to find another solution.

That being said, any change would mean lots of work on the players end so I don't intend to move hastily. Not even sure if there's a viable solution out there.

Also from what poster is saying, just simply having lots of chests in an area lags people's machines? If that's the case then nothing we do would change that situation. Unless twit wants to sponsor new machines for all players ^_^ get on that timber :-p

[PosterAnonymous Profile]

Also from what poster is saying, just simply having lots of chests in an area lags people's machines? If that's the case then nothing we do would change that situation.

Yep! A few players have been testing it for a while now (I remembered them talking about it on Mumble, hence the request for the result data). It is mostly dependent upon the client machine. Someone also provided me with the magic number that seems to pretty much kill your client as soon as you enter the area, which is good to know. I don't see any single player reaching that number realistically unless they're just hoarding every available block they encounter or they don't know how to put more than one item into a chest.

This idea would not have been an attempt to magically get rid of this problem, it was to be an attempt to shift the problem to an area that few use/enter and then link to it via Sign Shop. Any player entering that area, depending upon their PC, would be instantly hit with some lag. Since it's on-hold for now, it's nothing to worry about.

[TehRage Profile]

We need to make sure the topic of a new buy/sale plugin is discussed at the town hall. Best way to get lots of players voices and concerns addressed.

[thesillychicken Profile]

We need to make sure the topic of a new buy/sale plugin is discussed at the town hall. Best way to get lots of players voices and concerns addressed.

Submit topics to twitcrafttownhall@gmail.com

[Curley]

But as this conversation seems to be saying is the issue isn't sign shop. It's just chests. All plugins are gonna need a chest to store stuff.

[Goobler Profile]

A good place that's close to spawn and no one goes to is your old pub PA.

[Dio Profile]

But as this conversation seems to be saying is the issue isn't sign shop. It's just chests. All plugins are gonna need a chest to store stuff.

This is correct. The while the sign shop plugin may be inefficient server-side, it has nothing to do with the client side FPS lag that is caused by chests and signs. The client has to render each chest and sign, even if it's not being directly looked at.

I verified this with a download of the world map and I believe Vantes and TSC did a chest lag test in flatland. If you want to check it out it's around -400, 300.


Edit: Here is some data gathered from my tests. This was done by downloading a copy of the Flea Market and pasting it into a flat clean world. There were 2,456 chests and 1,115 signs in the Flea Market.

[DragonofHalo Profile]

Just a suggestion: if you need a location to build the repository, assuming that you decide to build it, try a decent distance from spawn and in a weird/undesirable location. Like, under an ocean. I can quickly get a Netherroad out there, so players will have about the same walk time as they would to get to the first location via overworld. And by quickly, I don't mean my normal "Under 6 months" quickly. I mean a few hours. Spread over a couple days
(Portal will have to be on Y greater than 20)

[PosterAnonymous Profile]

A good place that's close to spawn and no one goes to is your old pub PA.

Too heavily travelled foot-traffic wise.

Just a suggestion: if you need a location to build the repository, assuming that you decide to build it, try a decent distance from spawn and in a weird/undesirable location. Like, under an ocean.

Shut. Up. :evil eye:



EDIT: Not a big surprise, really. Anyone mildly observant should be able to tell I enjoy aquatic builds.

[DragonofHalo Profile]

Under, as in underground, since nobody else would build in the surrounding chunks.
But you could make a mini-storage based-aquadome if ya wanted

[abigail Profile]

what I did for my various stores (right before I made Cakes to Dye For a chain) was move my store storage from the basement of the mirage to my house in west leoville. not only just outside leoville but there are no apartments, high rises or 20 homes per chunk. So if i set up 50-70 chests it would not kill anyone's connection to walk through the area.

I don't think moving everyone's storage to one place would solve the problem. not that people should not use it. Like enterprising players that already have homes, and lots of the things (like 26 sign shop signs, which may not be exact but is close) could keep their chests in a place (not with the others) and people who were new could be free to (or pay if its not) to set up a couple of double chests and a few signs. Even 20 of people did that it could be less than a hundred chests.

it does not solve the sign rendering problem, but spreading chests over chunks or the whole server would bring the chest-lag down. if an area was only plagued by one or the other it would be better.

con: it took me just under an hour to link 26 signs, after i got used to the process (the first time i set up in it was around 2 hours, but that was the butterfly factor)
pro: i pretty much only have to do it once for each store, set it and forget it. And dio's tunnel significantly cut down the already short time it takes to get to that house. (it was less than a couple minutes to begin with, but shorter)