EVE General Discussion

Sun is up in 'murricah.
Log in time comes.
Some of the boyz will wait for the Logoffski Heroes with flowers and spiky cookies.
Hehe, we will have a wonderfull afternoon.

So you say, but I personally am hoping that TEST does not learn from today and will in another date, have their carriers left by PL to the mercies of CFC battleships fleets.

Well, unless the system gets remapped again, but SURELY it won't happen so fortuitously twice in a row, right?

I would expect CCP to "give a rats ass" about any battle or aspect of EVE. And I think they do.
They are dependant on us as customers, and as soon as they "dont give a **** " we will pick up on it and leave,
no matter how addicted we might be to the game.

When a customer of my company got gripes with our service we apologize and try as far as possible to reduce the impact.
The customer expects, as he/she should, to have service delivered as advertised. If we are unable to deliver, I should
accept the "sour grapes" Im served listening to the critizism.

The "aftermath" of critizism from a customer should be reviewing what went wrong and how to avoid it happening again.
If everyone that made a mistake in a company got axed at first offence, then you would run out of employees very quickly.
But at the same time a customer have every right to critizise if so inclined.

Its the service providers job to filter thru the noise to find useful feedback.

Why exactly does a node remap require disconnecting players at all?

Don't we (the game clients) connect to some sort of proxy servers anyway? And if so couldn't the sol server tell the proxies, "hey guys, we're freezing the simulation here, wait for handover" to signal a handover condition. Because as far as I understand it, a single python thread is the core of the simulation for one system. That server then could simply serialize all its objects, give them to the new node which de-serializes and thus recreates the state and then signal the proxies again that it will continue "over there on the new node" completing a handover sequence without disconnect, but maybe with freezing on all players for a few seconds which - under max TiDi is not really different anyway.

Yes simplicistic view and there's probably some technical limitation in there somewhere preventing it. But wouldn't that be - besides brain in a box - a very powerful and desirable functionality that would benefit not just single big fights but allow for dynamic on the spot automatic load balancing just as well as feature tests on live node via migration of state to different nodes?

I mean, the biggest growth problem EVE faces is the fact that a lot is single threaded. CPUs just won't get much faster. Everyone working in the server business knows this. But we have and will get more massive multi-cpu, multi-socket and multi-cores and that in the end means if you cannot make your code run multi-threaded you will hit performance barriers that much faster. Being able to dynamically split off at least that single-threaded smallest unit of processing would buy a lot of time until something more parallel or let's say per-grid threads could be implemented.

Anyway, don't really expect any answer but just throwing it out there as a TL;DR that implementing a node handover system without disconnects is something CCP should invest resources in.

Of cause it's a game but ccp provides a paid for service, so just because it's a game is no excuse to hide behind. Any live service related infrastructure needs safety systems and checklists when changes are made. It's clear ccp has neither of these on it's critical systems. As a QA manager who's company provides a service on a similar infrastructure I find ccps practices to be inadequate - these mistakes are easily avoided if risk assessments are done and simple protective measures are put in place.

Human error is easily avoidable, the fat fingered sys admin who made the mistake isn't to blame, it's his/her department manager who is because they did not mitigate the easily avoidable risk with a simple check list.

You continue to capitalize libel which is a great mystery to me. First, you have to understand that seriousness of intent is very important in determining if a statement is libelous. Second, we are speaking on a public forum owned, operated, and moderated by CCP. They have the right to censor anything we say without reason or justification. This alone means they have no grounds for claiming libel. Lastly, I don't think anyone seriously thinks Booda is hiding in the ventilation shafts in London waiting to hit the kill switch when a fleet fight goes south for Test.

Contrary to what you say, what is said on the forums, truth or nonsense, is far less important than what happened today in the actual game. You'll remember that EVE Online is almost certainly their sole source of revenue. When games are mismanaged consistently, they die. There's an elephant graveyard out there somewhere with all the names of triple a mmo's who died slow awful deaths because their developers failed to make a product worth paying for. When you say it doesn't affect the "stock market" you're dead wrong. They have a fiduciary responsibility to their stock holders to make money. And while I think this has all been a bit melodramatic, they lost some good faith today.

It gets.. confusing when you're in the middle of things. So yeah.

Plus, with TiDi, there's hardly 1000 thing needing your attention RIGHT NOW.

This is why assuming that a large mass of people will all fulfill rational expectations and Bayesian updating requirements will lead you to a terrible, terrible place.

This is why better people come up with doctrine fits, so we can all fly nonhorrible fits.

Maybe CCP is banking on being too big to fail.

I mean, it worked for the banks next door to it, right ...

Maybe you should have watched Boat on themittani.com stream, because he knew why.