TMA: moving a server cluster

[tgr]

These fellas found the safer paths to be, indeed, safer:
http://bl3h-united.com/?a=kill_related&kll_id=177667

Don't forget the BPO/BPCs getting the same icon as the first BPO/BPC (might just be BPC, not sure. I can't be arsed to check).

I'm amazed at all the stupid mistakes they let slip in through not just large patches, but small (supposedly much more focused) patches as well. It's fucking clownshoes, is what it is.

[Reg]

The autopilot bug where it resets to shortest path from prefer safer happens frequently after a patch. It's one of those things you learn to expect along with having to redo your overview.

[tgr]

The autopilot bug where it resets to shortest path from prefer safer happens frequently after a patch. It's one of those things you learn to expect along with having to redo your overview.

You mean that the default choice of "prefer safer" is then set to "prefer shortest"?

I was under the impression that it preferred the shortest route even if you specifically went in and chose "prefer safer route".

[lac]

Tyrannis 1.0.3 patch to be deployed on July 1, 2010

FIXES

The autopilot was mistakenly providing the shortest route to your destination even when you had the prefer safer option selected. We sent it back to flight school and the safer option is again working correctly.

Some blueprint icons were being displayed incorrectly based on their type. This has been corrected.

An error was being generated while trying to access items in a corp hanger while in space.

You are now able to trash items normally that are located in a corporate hangar array at a POS.

The Planetary Commodities cargo hold on the Primae couldn’t be loaded with Water or Oxygen. It can now.

An error was being generated why you tried to edit more than one contact at the same time. This now works as intended.

[Reg]

Oh wow. That is a new one. I thought you were talking about the old one where your autopilot setting just gets reset.

[lac]

Yea, they managed to have one of their annoying patch day resets turn into a full blown bug. I wonder if they'll reimburse those popped freighters.

[tgr]

Of course they'll reimbursahahahahahahaha no they won't.

[lac]

It wasn't me, it was the SAN. This devblog explains the drama that took place on 23/07/10 in the air-conditioned cold of the CCP server room in London.

[tgr]

So they can provide details on this, but the lag monster? Heeeeell no.

I seriously think everyone in 0.0 needs to start thunderdoming the larger trading hubs for a few weeks, just to make the lag problem visible to even those in empire. I know I had absolutely no idea how bad it was pre-stackless IO, because I was just faffing about in empire and not seeing any problems at all.

[Reg]

Heh you actually think CCP would pay more attention to 0.0 problems if you upset a bunch of empire dwellers who are playing the game all wrong?

[tgr]

I expect that one of two things would happen.

1) They fix lag.
2) They ban all those who are "griefing" (aka using the wardec mechanic as intended to bring the fighting into hisec).

Probably 2).

[Reg]

Lag is their biggest problem. It's what's spoiling the game for people that are playing the way they designed it to be played. Do you really think they're purposely refusing to fix it for some reason?

I'm no big fan of CCP but I really can't believe that they're quite that bad.

[tgr]

It's not a problem for everyone who's playing the game the way it was designed, only for those who are part of large (for decreasingly smaller values of 'large') fleet fights. It's probably also affecting FW, but I've no idea how many people are actually still playing FW, but manufacturers, BPO copiers, mission runners etc are probably not seeing that much of these problems. Maybe they get traffic notification on a gate once in a while, but apart from that, to them eve works just fine.

I wouldn't be surprised if at least one guy is trying to figure out what the fuck, but when was the last time you saw any information from CCP in any way, shape or form on lag? All I've seen is how they've kept the node alive when CVA got fucked, and how the ships were not reimbursed by GMs in 6nj. That, and sisi tests which doesn't really seem to amount to much.

There's no information, there seems to be no work done, no progress at all, and it just looks like (from all appearances, since again, there's no information) CCP is ignoring this and hoping it'll go away.

I'm under no illusion that thunderdoming various trading hotspots in empire (i.e. 0.0 dwellers wardec eachother and run around in empire having 200v200 fights, so it's 100% legal) would even do anything to make CCP prioritize fixing lag more, but it's a damn tempting thought.

No, I guess I'll just have to live with the fact that the servers are in this state, and will continue to be in this state for probably a few years, and just move over to using covert ships to fag up enemy systems, since small gangs seem to be the only thing that's even reliable these days.

[Sir T]

Yeah I heard about that "Titans that jumped out 2 hours before they were killed in the system they jumped from reappeared in the system they jumped to the next downstime when the server rebuilt itself." bullshit yesterday. I mean what the hell.

The long term thing is that eventually it will only be small gangs of supercaps that will be able to move in 0.0, effectively making them invincible, again.

The wholeway the server is worked has always been strange. I remember way back during the BOB/Ascn war that we had massive traffic controls in the region I lived in (Solitude) The theory was that the systems are shared between different regions, and what goes on in one system could effect what goes on in seemingly random systems all over Eve. One of the benefits of stackless IO was that didn't happen anymore.

At the time we were all mad as hell as CCP were obviously sucking power away from other systems so that BOB could have fun. The fact that WE, mere low sec dewllers, were trying to have a war as well and we couldn't even move didn't seem to matter to them. Hell I remember getting Multiple traffic controls in Ankee when there was literally no-one in system. But Ankee was obviously tied to so,e region down south and THEY needed the juice to get BOB back first after a node crash or something.

[Stabs]

when was the last time you saw any information from CCP in any way, shape or form on lag?

Patch notes for Tyrannis 1.0.2, released June 29, 2010

Need for Speed

    * Memory usage has been improved with missiles, maps, and PIN’s.
    * Frames per second has been improved when using warp disruption bubbles.

# Players will no longer see the fleet list scroll to the top in the flat list, when a member enters or leaves a large fleet.
# The overview and brackets have been optimized and will flicker less when in a big fleet.
# The overview window refresh rate has been optimized.

Patch notes for Tyrannis 1.0.1, released 02 June 2010

Optimizations were made to the Planet view. It will now load faster and run smoother.

Patch notes for EVE Online: Tyrannis, released Wednesday 26 May 2010.

Need for Speed

    * Made various improvements to the EVE server software and corified more of its components along the way.

and of course the moving of the server cluster.

[Sir T]

Adn they have been so successful 

[lac]

CCP knows exactly where the lag is.

They have competent database people and it's real easy for them to tell what's causing the lockups. Any 25$ monitoring app could provide the necessary data, really.
Seeing where stuff locks up is the easy part. Modifying your code to not make stuff fuck up is a lot harder.

Those SiSi tests they are doing are bullshit. It's Pr designed to keep both their customer support team and the concerned part of their clientèle occupied and feeling useful.

Running a comprehensive log of a node were is a fleet fight is going on, with the hardware they have, won't really interfere that much with the experience of the players. When the node is going tits up anyway, that 10% loss of performance caused by the logging every little detail won't change anything.

Yet it shows so much. Eve is crippling on it's code. And that means it's pretty much done. Sure, they can tag shit on, and that seems to be what they are doing, but unless somebody takes this whole project and redesigns it from the ground up we'll be left dangling.

[Sir T]

A rebuild would mean taking the servers offline for a while. Hell it might even mean deleting some skillz. And they are afraid of doing that as if people did not get their fix they might never come back.

It pays them more to keep people addicted.

[Stabs]

The idea of them taking the entire game apart and then trying to put it back together again makes me nervous for some reason.

[Endie]

The idea of them taking the entire game apart and then trying to put it back together again makes me nervous for some reason.

"Well, there we go, that's her rebuilt."  [Dusts-off hands, wipes on oily rag]
"Um, guys... should this bit have gone somewhere?"

[tgr]

when was the last time you saw any information from CCP in any way, shape or form on lag?

Patch notes for Tyrannis 1.0.2, released June 29, 2010

Need for Speed

    * Memory usage has been improved with missiles, maps, and PIN’s.
    * Frames per second has been improved when using warp disruption bubbles.

# Players will no longer see the fleet list scroll to the top in the flat list, when a member enters or leaves a large fleet.
# The overview and brackets have been optimized and will flicker less when in a big fleet.
# The overview window refresh rate has been optimized.

Patch notes for Tyrannis 1.0.1, released 02 June 2010

Optimizations were made to the Planet view. It will now load faster and run smoother.

Patch notes for EVE Online: Tyrannis, released Wednesday 26 May 2010.

Need for Speed

    * Made various improvements to the EVE server software and corified more of its components along the way.

and of course the moving of the server cluster.

And none of these really touch on lag, grid-load or session change issues. The server cluster moving had absolutely nothing to do with fixing any of the problems I'm thinking about. All I see are hugely annoying (but somewhat superficial, unless you target the wrong guy because the damn overview keeps reorganizing itself) issues that also came along with Dominion.

I just wish they'd figure out and fix whatever it is that's ruining the eve big-battles experience, but until then I'll just go about and changing the way I play the game. I've been skilling towards hounds, and I'm thinking of going down the bomber/gsrecon route instead. I've never really done that kind of playing. I've been an industrialist, and a low-level fleet fight grunt, but I've never really done this, so I guess I might as well give it a try. It's a big sandbox after all, might as well have a go at it.

Actually, I'm starting to wonder if lag and gridload isn't that much of a problem, it's the whole changing session to a different system and/or physical node that's the really fucked up part. How many times haven't people mentioned that once you get in system, moving about is fine? That makes me wonder just how much work it might take to fix it, but they haven't really said anything about what the core problem is, if they've found it, and how much time they estimate it'll take to fix it so all we have to go on is ... absolutely nothing, which is the most frustrating bit. If they did know where the lag is, they should say so, so we'd stop wondering. It's the wondering part that's the most frustrating part, to me. That, and the inconsistent crap that follows.

[Stabs]

I'm not terribly technical but my understanding that lag is a broad term encompassing a wide range of different issues that cause poor client-server communication. Fixing missile graphics or whatever won't magic wand lag away but lots of little tweaks like that should help.

Also why won't the new more powerful computers improve things?

[TheDreamr]

Also why won't the new more powerful computers improve things?

Because you can't polish a turd.  Throwing more power at a non-specific performance problem only works up until a certain point, then it becomes prohibitively expensive to mask inefficient and poorly optimised code by just buying faster hardware to run it on.

If you can identify a specific problem then a combination of optimising code and throwing the correct hardware at it becomes much more realistic and cost-effective.

[Reg]

The the big fleet lag only appeared a couple of major patches ago though didn't it? If that's the case it seems to me that the problem should be isolated to the changes made during that patch rather than it being caused by general "shitty" code. That's why I don't understand why it hasn't been fixed already.  Is it possible that CCP has their best network coder off working on some other game - or maybe he left the company?

[slog]

Why do so many folks assume it's a network problem?

[Phildo]

Lag and various black-screening issues have existed for as long as the game itself has.  What Reg is thinking of is that it was significantly improved upon in Apocrypha (or thereabouts) and we had a few months where it was possible to fit up to a 1,000 people in a system.  Then Dominion happened and the servers basically went back to how they were before Apocrypha, with slightly altered behavior.