I just got disconnected with that weird "Java exception socket" thingy message also before I lagged/disonnected out I noticed that everyone had rollback so for example, If you had a stack of gold, died and then relogged you'd get the gold back.

Just pointing this out as there may well be people taking advantage of this and duping

Rollbacks are universal so you can't actually get more than what you had but it does fix losses due to things like death. Also players would have to be able to predict when rollbacks will happen to determine when they should put away gold, which is highly unlikely. (If anyone has figured out how to predict it please contact sawine since that information would be crucial to preventing rollbacks in the future)

Well I kinda now when it's when the server is saving it takes about 5sec lagg but the rollback its about 15 secs before it kicks you out dunno if that helps but here (although I still lost my items and in the previous rollback <object class="emojione" data="https://resources.enjin.com/1489581540/themes/core/images/emojione/svg/2639.svg?0" type="image/svg+xml" standby=":(">:(</object>)

Usually you can tell when all the guests spamming the chat for help suddenly stop :'P

Currently scanning every chunk of the server to find the corrupted ones.

Take your time, sawine. Better safe, the sorry, right?

Could the source of this error come from loading non explored chunks that are surrounded by 1.7 ones?

Beause the error started about half an hour after I explored a non explored chunk near my city.

Um... It might be my fault. I'm being a wool whore last night, and I sort of just ran around in the unexplored world (1.8 updates, out at around 6000 x, 62 y, and 4000 z. I sort of just grab all the resources I deem valuable or critical to my projects. That loads A LOT of crap on the server, and I KNOW that what is loaded doesn't go away... so... yeah... that was around like 5 am PDT. I'm sorry if it's my fault.

Also, Sawine... roll backs are impossible to predict. You'd have to keep track of EVERYONE and tell them (don't go beyond this point) and pretty much just reset all the map that no one utilizes, and tell people to stay within the set parameters. This doesn't eliminate rollbacks, but lowers the frequency at which they happen (YES!!!! STAT CLASS PAYS OFF... I hate university, so many long words lol).

With the HUGE amount of guests getting on, it's putting a huge strain on the hardware on the server. From what I know, the more people on a server, the more information is transferred. With an already F*ed up server (no offense, but I've been hearing from people that the server has been F*ed up for months, I've only been on for about a week on this server ), we'd just pretty much have to wait for the new server.

I will suggest however, make the server a temp white-list only as this will decrease all lag and reduce the frequency of rollbacks (because gw is pretty much a huge sector that doesn't NEED to be used right now). And then when the new server comes out, turn white-list only off and return to normal operation. That is just my two cents, Sawine.

Where are your Servers located? I'm good with hardware. I'm in Southern California, so if it is located near I can take a look at the hardware and see if there is a problem there. You COULD have a capacitor problem (which is quite common, actually).

~John (dunsjohn)

Hey,

Ok... so what happened is that our world is on a "partitioned" area of the RAM to make it act as a disk to store data (RAM disk).

The amount of space allocated to the map was 4gb. So what happened is that the map expanded beyond that size limit and the last chunks couldn't be rendered correctly so they corrupted and bugged the server.

We should be moving the world to the new hardware in the next hours if everything goes as expected, so those kind of issues should not happen again.

Cya

Hey Sawine,

How much RAM do you expect to have? I know that a recommended is around 16-32 just for a minecraft server. What type of hardware are you using, I have loads of spare RAM I could mail off.