These forums have been archived and are now read-only.

The new forums are live and can be found at https://forums.eveonline.com/

EVE General Discussion

 
  • Topic is locked indefinitely.
Previous page12
 

Credit where credit is due: Team Gridlock shout out

First post
Author
Spurty
#21 - 2011-09-25 21:45:43 UTC
Malcanis wrote:
Well there was a fun little scuffle in Great Wildlands, with dozens of supercarriers and many titans, fighter and fighter-bombers all over the field, 300+ in local and zero lag. When I say zero lag, I mean absolutely no perceptible lag effects at all. There might as well have been 3 in local, not 300.

Whatever other problems EVE has right now, TQ performance isn't one of them.

Well done and a big thank you to Team Gridlock for helping to make this afternoon's fun possible.


http://tinyurl.com/3bz83n5


FY-URL (So EVE-FORUMS doesn't sterilize it)

There are good ships,

And wood ships,

And ships that sail the sea

But the best ships are Spaceships

Built by CCP

Malcanis
Vanishing Point.
The Initiative.
#22 - 2011-09-25 21:53:16 UTC
CCP Veritas wrote:
I was monitoring the server during that fight, and have had a preliminary look at the server logs for it, and there were definitely problems for some folks in regards to jumping between systems on the server side. I'll need to dig into it more in the morning.


Of particular interest is why Player A gets lagmonstered hard while Player B in the same fleet on the same grid sees minimal effects.

Obviously any hints that you can give us on how to affect this on the client side would be extremely valuable. I had brackets off, all effects off, ship, gate & planet only overview, if that helps.

"Just remember later that I warned against any change to jump ranges or fatigue. You earned whats coming."

Grath Telkin, 11.10.2016

Malcanis
Vanishing Point.
The Initiative.
#23 - 2011-09-25 21:53:53 UTC
Spurty wrote:
Malcanis wrote:
Well there was a fun little scuffle in Great Wildlands, with dozens of supercarriers and many titans, fighter and fighter-bombers all over the field, 300+ in local and zero lag. When I say zero lag, I mean absolutely no perceptible lag effects at all. There might as well have been 3 in local, not 300.

Whatever other problems EVE has right now, TQ performance isn't one of them.

Well done and a big thank you to Team Gridlock for helping to make this afternoon's fun possible.


http://tinyurl.com/3bz83n5


FY-URL (So EVE-FORUMS doesn't sterilize it)


tyvm, Mr Spurty.

"Just remember later that I warned against any change to jump ranges or fatigue. You earned whats coming."

Grath Telkin, 11.10.2016

Alexandra Alt
Doomheim
#24 - 2011-09-25 22:08:40 UTC
Malcanis wrote:
CCP Veritas wrote:
I was monitoring the server during that fight, and have had a preliminary look at the server logs for it, and there were definitely problems for some folks in regards to jumping between systems on the server side. I'll need to dig into it more in the morning.


Of particular interest is why Player A gets lagmonstered hard while Player B in the same fleet on the same grid sees minimal effects.

Obviously any hints that you can give us on how to affect this on the client side would be extremely valuable. I had brackets off, all effects off, ship, gate & planet only overview, if that helps.


That might be aplicabe to the fight itself yes, but how about all those left behing during the travel to said system being picked on while trying to log back in and stuck in black loading screens and progress bar being shot helpless ?

I'm pretty sure no setting/config or PC setup would fix that.
beor oranes
Annihilate.
#25 - 2011-09-25 22:25:32 UTC
You are probably right Alexandra Alt, there is probably some problems server side that cause people to get stuck. I was just offering to anecdotal evidence of my own to suggest that not all lag is caused by the server but could be affected from the client side as well. I have myself suffered random black screens when jumping between two empty systems and know how annoying it is, I have also suffered this and my mate didn't on the same gate at the same time, so I would think that some of the problem must come from my end it cant be all server end. Anything that will help in the setup of your computer to alleviate part of the problem (if there is one there) is surely a good thing to know, right?
Alexandra Alt
Doomheim
#26 - 2011-09-25 23:14:33 UTC  |  Edited by: Alexandra Alt
beor oranes wrote:
You are probably right Alexandra Alt, there is probably some problems server side that cause people to get stuck. I was just offering to anecdotal evidence of my own to suggest that not all lag is caused by the server but could be affected from the client side as well. I have myself suffered random black screens when jumping between two empty systems and know how annoying it is, I have also suffered this and my mate didn't on the same gate at the same time, so I would think that some of the problem must come from my end it cant be all server end. Anything that will help in the setup of your computer to alleviate part of the problem (if there is one there) is surely a good thing to know, right?


Please excuse me as I'm quite uppset, aimed for a pleasant pew pew afternoon today when I could have some time outside with mrs and a few friends, but instead I was hungry for some lasors and ended up stuck in a black screen being killed by a handfull of noobs not being able to do anything about it, having to pod myself to get back.

Regarding the issue at hand I find it quite difficult for the simple and quick check reason that my alt, had no issues whatsoever in empire, therefore could not be an issue on my end, same pc, same network connection, same variables on my side, nevertheless you are right many issues might not be eve server related, hell, might not even be the player/player connection fault, the routing can go haywire sometimes and in those circumstances I'll be the first one to point it out, as fast as I criticize CCP for an issue that arise I believe to be their problem, I'll also thank when it is deserved, and for that, I have to thank as CarbonIO is actually working amazingly great as tested this thursday in Sisi.

The issue with the black screen has been plaguing players from long time already and imo as a developer there could have been a bandaid fix to avoid it for quite a while already, I don't know the intricates of CCP server/client communication though, but I would guess that all it would need would be simply waiting for the whole client to be fully loaded and responsive and send a packet to the server allowing for the warp initiation, something that doesn't seem to be happening, the 'green' light sent to server to initiate the warp is happening way before all resources are loaded, friends had half ui loaded, no chat whatsoever nor overview, only ship and modules, unable to do anything, all frozen, while being shot by reds @ the gate (and this in a system with about 15 ppl), my guessing is, server load was tremendous, queueing instructions, and the client wasn't in synch with the server.

Again, imho, this things aren't hard to fix, obviously I'm talking without knowing how the server works, probably the server has some screwed up threading design that allows for this easy desynch that CarbonIO is about to fix, but I had to rant, I got really pissed, and it got me even more uppset because I don't like at all to not have control of what's happening, it feels really bad to know you're being shot at, in a blank screen, where u know that if you quit client won't do a thing, and staying will just keep you staring at a black screen.
CCP Veritas
C C P
C C P Alliance
#27 - 2011-09-25 23:33:51 UTC
These black-screening issues usually trace back to problems with the session-change architecture on the server side. We've deployed a slew of fixes for various ways it could go haywire over the past year and change. It's unfortunately not surprising to me that it still falls on its face under load though, it's pretty gnarly.

CCP Veritas - Technical Director - EVE Online

beor oranes
Annihilate.
#28 - 2011-09-25 23:38:53 UTC  |  Edited by: beor oranes
I sympathise mate, I really do. I have had it happen to me on a number of occasions and I can see why you are annoyed. All I was suggesting was that maybe there was more to it than just 'a server screw up'. However, I am open to the idea that it could something that is a problem at the server side and nothing that a player can do to improve or worsen. I must confess that I am not technical to really understand how it works but I would have thought that if the solution was so simple then why hasn't it been done?! Clearly there is something else happening whether that be client or server or somewhere in between that causes this problem and CCP haven't managed to narrow it down enough to stop it happening so frequently. Maybe CCP Veritas can give us a better insight into why it is happening and whether such a fix would work if it could be implemented?

Edit: He beat me too it. Thanks CCP Veritas for explaining a bit more about the reason for black screens.
Alexandra Alt
Doomheim
#29 - 2011-09-26 00:03:53 UTC  |  Edited by: Alexandra Alt
CCP Veritas wrote:
These black-screening issues usually trace back to problems with the session-change architecture on the server side. We've deployed a slew of fixes for various ways it could go haywire over the past year and change. It's unfortunately not surprising to me that it still falls on its face under load though, it's pretty gnarly.


Does CarbonIO fix the session-change architecture is it a threading problem?

I am the first (probably) to criticize rush deployments, but I would classify it's deployment as urgent tbh, winter is comming, the load will increase.

Out of curiosity, is CarbonIO already developed in a compiled language and not python ? (Sorry, I'm kinda python hater, not because of EvE mind you :P)

edit:

I just re-read the CarbonIO blog posted by CCP Curt, basically my initial assumption (the lack of horizontal scalability for Python) was correct, it's really a pain when you want to take a truly multi-threaded implementation and you are limited by the interpreter itself.

I do understand python allows for quite faster iteration, though I'm guessing you guys already reached the conclusion that stepping away from it is the best solution, I can only imagine all this years ago, who would have thought 640kb was not going to be enough after all :)
Rakshasa Taisab
Sane Industries Inc.
#30 - 2011-09-26 02:09:30 UTC
No Lag, Fun Features and Exciting Future Plans.

Pick any one.

Nyan

Malcanis
Vanishing Point.
The Initiative.
#31 - 2011-09-26 06:22:43 UTC
CCP Veritas wrote:
These black-screening issues usually trace back to problems with the session-change architecture on the server side. We've deployed a slew of fixes for various ways it could go haywire over the past year and change. It's unfortunately not surprising to me that it still falls on its face under load though, it's pretty gnarly.


I can understand that when the node has to grind through 850 warp commands or whatever, some guy is going to get the sticky end of the stick and his warp will be executed last; what I've never really understood is why the server keeps picking on the same guys to get said sticky end. Like once it's "chosen" Player A to get the lag shaft, he gets lagged the whole fight, whilst Player B continues to receive favourable treatment.

Levelling out this inequity is one of the biggest attractions of Time Dilation, btw. It wouldn't just reduce the frustration caused by unfairness - a fleet is often constrained to act at the speed of its slowest members.

"Just remember later that I warned against any change to jump ranges or fatigue. You earned whats coming."

Grath Telkin, 11.10.2016

Headerman
Native Freshfood
Minmatar Republic
#32 - 2011-09-26 06:51:12 UTC
Sounds like Session Changes is due for a major upgrade next, ala Carbon IO?

Australian Fanfest Event https://forums.eveonline.com/default.aspx?g=posts&find=unread&t=90062

Vincent Athena
Photosynth
#33 - 2011-09-26 15:19:18 UTC
CCP Veritas wrote:
These black-screening issues usually trace back to problems with the session-change architecture on the server side. We've deployed a slew of fixes for various ways it could go haywire over the past year and change. It's unfortunately not surprising to me that it still falls on its face under load though, it's pretty gnarly.


Will TiDi allow it to not fall on its face? Also as long as someone brought up TiDi...

Do we get to test it this week?
Are solar all the solar systems being run by one node grouped together on the map, or will be get odd things like Tar having TiDi because of a big battle in Null?

Know a Frozen fan? Check this out

Frozen fanfiction

CCP Veritas
C C P
C C P Alliance
#34 - 2011-09-26 19:36:15 UTC
Alexandra Alt wrote:
Does CarbonIO fix the session-change architecture is it a threading problem?

No. CarbonIO has been running on the server for some time now, so I feel comfortable saying it doesn't solve the problem.

Alexandra Alt wrote:
Out of curiosity, is CarbonIO already developed in a compiled language and not python ? (Sorry, I'm kinda python hater, not because of EvE mind you :P)

Yes, it's in C++.

Headerman wrote:
Sounds like Session Changes is due for a major upgrade next

It's been on various teams' wish list since I've been at CCP (a year and a half-ish), but it's one of those big-scary-might-break-everything things that somehow keeps being put off Ugh

Vincent Athena wrote:
Will TiDi allow it to not fall on its face?

It should help, though I'm not certain quite how much, since I'm not certain what the exact problem we're dealing with here is. Depending on how aggressively we tune TiDi to kick in once load is laid on, it might not respond fast enough. The downside of making it aggressive enough to kick in before such things can go down is that it might kick in when it really shouldn't. It's a balancing act we'll need to play with once we have real life datapoints to track.

Vincent Athena wrote:
Do we get to test it this week?

I hope so; our test plan is quite behind schedule already ><

CCP Veritas - Technical Director - EVE Online

CCP Veritas
C C P
C C P Alliance
#35 - 2011-09-26 19:36:29 UTC
Vincent Athena wrote:
Are solar all the solar systems being run by one node grouped together on the map, or will be get odd things like Tar having TiDi because of a big battle in Null?


Yes, they're almost always grouped by constellation and region. There's also a hard boundary between null, empire and wormhole nodes, so bleeding through type of space is extremely unlikely.

CCP Veritas - Technical Director - EVE Online

Andy Landen
Deep Core Mining Inc.
Caldari State
#36 - 2011-09-28 21:09:01 UTC
Thanks for the update to my thread: post 114794.
Let's get some dates and commitments now please.

"We cannot solve our problems with the same thinking we used when we created them." Albert Einstein 

Jita Alt666
#37 - 2011-09-28 22:14:15 UTC
Quote:
It's been on various teams' wish list since I've been at CCP (a year and a half-ish), but it's one of those big-scary-might-break-everything things that somehow keeps being put off


Please make it a priority. Everything regarding FIS comes down to fair ability of pilots to conduct their actions in any possible fleet setting.

I do have to agree with the OP: Things have been improving, a fight of 200-400 is no longer a lag grind over 700 in local though and things are still pretty gnarly...
Taleris Kline
Pator Tech School
Minmatar Republic
#38 - 2011-09-29 12:30:46 UTC
CCP Veritas wrote:
[
Headerman wrote:
Sounds like Session Changes is due for a major upgrade next

It's been on various teams' wish list since I've been at CCP (a year and a half-ish), but it's one of those big-scary-might-break-everything things that somehow keeps being put off Ugh


Are we talking undocumented code that hasn't been touched in 5 years gnarly or undocumented, obfuscated, spaghetti code that was written by a drunken monkey gnarly?



Previous page12