Server Issues: Crashes, Lag, Restarts, Rollbacks

[DeadlyPencil]

you should stop new account registrations if you cant handle the current load.

[Caliku]

241 people.

Awesome to see the numbers get so high. It's a valuable server stressing situation and freezing registrations would stop that feedback from coming in.

It may suck, but it's entirely necessary.

[DeadlyPencil]

241 people.

Awesome to see the numbers get so high. It's a valuable server stressing situation and freezing registrations would stop that feedback from coming in.

It may suck, but it's entirely necessary.

you weren't on in the morning obviously.... the server crashed due to the amount of people online. this isn't some production game where the servers they have are the ones they are going to end with. they dont need to stress the servers. whats the point of stressing a server when you know it cant handle the load other than causing it to crash for no reason.

[Caliku]

241 people.

Awesome to see the numbers get so high. It's a valuable server stressing situation and freezing registrations would stop that feedback from coming in.

It may suck, but it's entirely necessary.

you weren't on in the morning obviously.... the server crashed due to the amount of people online. this isn't some production game where the servers they have are the ones they are going to end with. they dont need to stress the servers. whats the point of stressing a server when you know it cant handle the load other than causing it to crash for no reason.

loftar has already stated that it's not the server. It's an issue with the code and he's working on it. He just has to figure out how to fix it, if I recall correctly.

[DeadlyPencil]

241 people.

Awesome to see the numbers get so high. It's a valuable server stressing situation and freezing registrations would stop that feedback from coming in.

It may suck, but it's entirely necessary.

you weren't on in the morning obviously.... the server crashed due to the amount of people online. this isn't some production game where the servers they have are the ones they are going to end with. they dont need to stress the servers. whats the point of stressing a server when you know it cant handle the load other than causing it to crash for no reason.

loftar has already stated that it's not the server. It's an issue with the code and he's working on it. He just has to figure out how to fix it, if I recall correctly.

you haven't even read this topic have you? you just replied to my post without reading anyhting.

let me summarize.
-lothar found the problem that was causing the lag, it was the bandwidth shaper thingy, it is disabled now. and not the cause of the current lag
-lothars post above my post explained that jorbs connection cannot handle the current load.

[Caliku]

Why don't you stop bitching and donate some cash for a better connection or dedicated hosting. That shit ain't cheap you know.

I will fucking murder you

[DeadlyPencil]

Why don't you stop bitching and donate some cash for a better connection or dedicated hosting. That shit ain't cheap you know.

I will fucking murder you

RAAAGEEEEEEEEEE

[loftar]

RAAAGEEEEEEEEEE

To clear things up slightly, the server hasn't crashed because of load; the crash this morning was just because of bugs in the new code we added yesterday. They have been fixed now. Stress testing is still very valuable, since I can observe just what happens when the server is under load (how the VM* system behaves, what I/O bottlenecks I've got, how the CPU usage scales with the load, where bandwidth is spent and lots of other things). As for load limitations -- if we do anything, it's probably wiser to put a limit on the number of users logged in rather than stopping registrations. Then, at least people can register accounts and participate in the forum or try again later. I'm not sure whether we'll be doing anything, though -- the number of users logged in seems to balance itself out anyway since people just stop playing when it gets bad enough.

Also, I did not say with definitiveness that Jorb's connection cannot handle the load. I did add that I'm not sure whether the lag is because of that or because of bad packet scheduling; that it may become better if I just get the bandwidth shaper into shape (hurr hurr, pun intended); and that I'll withhold judgement on the issue until then. I have also found out since that I can probably improve the packet retransmission code in the server quite a lot, which may also delay the eventual breakdown of Jorb's connection by quite some time.

* That's "virtual memory", not "virtual machine", for any of you newfags out there.

[Drowner]

$10 bucks NOW or someone much scarier than me will hunt you down and gut you!
https://www.paypal.com/se/cgi-bin/websc ... a5ce177198

[Tower]

Server crashed again.

Thats too bad, I was just building a second log cabin too.