| Author |
Thread Statistics | Show CCP posts - 12 post(s) |
|
Wrangler
   |
Posted - 2006.09.16 22:35:00 -
[1]
As most of you have noticed, we have had problems with Tranquility in the form of lag, nodes dropping and sometimes the whole cluster going down.
To explain this we have a whole blog, everything isn't great, but things will improve and there is definately light at the end of the tunnel.
Wrangler
Assistant Community Manager
EVE Online
Contact Support - Contact Moderators - Report Bug - Submit News Leads
Knowledge Base - Player Guide - Policies - Join ISD - Fan Submissions |
|
|
Wrangler
|
Posted - 2006.09.16 23:15:00 -
[2]
 Originally by: ArchenTheGreat
"But as stated earlier, Dragon solved a lot of issues. We're seeing better performance on the servers and the client, a boatload of issues were fixed as we can see from our internal error log counters, but there is still some way to go."
I am seeing it completely different. I have lower FPS (client lag is bigger), I have worse server lag (modules activate time went to 20-30s), I have lots of stability issues (client locks up during jumping when two clients are active) etc.
I had no such issues before patch. I have no idea how you implement your error counters but maybe instead of this you should try to PLAY this game sometimes. I can get you on lvl4 mission in Tamo with me to show you how bad it is.
I'm sorry, but you are taking that out of context. The full blog explains that while dragon improved things, it also messed a bit with other things, which is why you don't see an improvement. Once we have those issues sorted out, you will notice the improvements. But this is just the *really* short version, you need to take in the whole blog. 
Wrangler
Assistant Community Manager
EVE Online
Contact Support - Contact Moderators - Report Bug - Submit News Leads
Knowledge Base - Player Guide - Policies - Join ISD - Fan Submissions |
|
|
Oveur
|
Posted - 2006.09.17 14:24:00 -
[3]
Originally by: Jim McGregor
Nice blog. :) So a node can only hold the solarsystems from the same constellation? I thought a node could handle solarsystems from several different constellations...
Reply to this and ill spread the word in the forums, when we talk about load balancing. :)
We usually group solar systems by constellation as default, but individual solar systems are regularly pulled out of their constellation and distributed seperately, so a node can hold any mix of the above.
Senior Producer
EVE Online
|
|
|
Oveur
|
Posted - 2006.09.17 14:27:00 -
[4]
Originally by: Decairn
There are 3rd party tools to find memory leaks. Are those used at all or is the language causing the leaks not supported?
We have lots of them, but occasionally we get leaks which elude us. Especially since many of these tools can't be easily deployed to a live production environment like TQ.
Senior Producer
EVE Online
|
|
|
Oveur
|
Posted - 2006.09.17 14:28:00 -
[5]
Originally by: Ralitge boyter
Good to see you guys are on the case.
Now the only thing I am wondering is how can the two systems (China and World) be so diffrent...
Would it not be wise to try and push to having the same codebase? Just wondering of course. Though I de recall a certain software developer ehm whats their name again, Microsoft or something doing the same with their products due to porting issues they had during development.
They claimed it would cut cost and save time in development of much needed new things and improvments.
Ah, never mind they must be crazy. They probably stoped existing by now.
They have the same codebase. Dragon got it's name from being the first release to China.
Senior Producer
EVE Online
|
|
|
Oveur
|
Posted - 2006.09.17 14:36:00 -
[6]
Originally by: Audrea
Thanks for keeping us update, though I still think EVE needs more devs working on fixes, than new stuff :)
Other than that, there is one thing I would like to get clarified:
Originally by: Oveur
We're also seeing an issue with ships within gangs being able to lock down a single solar system.
Does that mean you want to change the way alliances currently lock down solar systems in preparation for dread assaults?
I just want to know, as such change (meaning allowing fleet to jump in and not suicide by doing that) would significantly change the tactics of POS warfare.
This hasn't got anything to do with gameplay, this is a bug where nobody within the solar system can move at all. Remove the ship and everything goes back to normal.
Senior Producer
EVE Online
|
|
|
Sharkbait
|
Posted - 2006.09.17 15:11:00 -
[7]
i want to look at the tower, about to contact you in game. not wise you mention the location on the forums
Spank You later |
|
|
Sharkbait
|
Posted - 2006.09.17 15:34:00 -
[8]
i have the server logs with me now, i have mailed the programmers and this issue is being looked into right now.
i will try to keep you posted regarding it
Spank You later |
|
|
Sharkbait
|
Posted - 2006.09.18 21:02:00 -
[9]
just letting you know, we have 3 possible fixes for the current problems. we will be testing them on sisi over the next few days and hope to get them to TQ asap. this also includes a fix for the Tower problem that papasmurf fixed this afternoon.
altho we think we have fixes, the problems are still being looked into, we are also looking into the way the server uses it's ram in hopes of improving it as much as possible.
again, i will keep you posted as i get information
Spank You later |
|
|
Sharkbait
|
Posted - 2006.09.18 22:20:00 -
[10]
if you do find a tower with no forcefield, please don't kill it. just leave it alone and report the issue to a GM via the petition system.
Spank You later |
|
|
Sharkbait
|
Posted - 2006.09.19 14:11:00 -
[11]
there was a problem with the startup scripts. this wasn't anything to do with the hotfixes, but they have been fixed anyways.
i will keep you updated
Spank You later |
|
|
Sharkbait
|
Posted - 2006.09.20 09:48:00 -
[12]
Originally by: Miss Overlord
so whats the next step sharky if u can inform us
hmm. good question. the startup scripts was the problem with the hotfix. it's something that should have popped up sooner but for some strange reason didn't. the hotfix with the debugging stuff was put out, but there was a problem with the memory debugging feature that was added. not sure what the problem was, i just know it was working fine on a local host and the 5 days it was on sisi before it went to TQ. so i guess the next step is for Porkbelly to fix that. Once he has fixed that he will be looking into the memory problems.
Richard thought the fix for the pos problem most probably wouldn't work, so Papasmurf roled the fix back (from sisi, it never made TQ) and started working on it again yesterday morning. he got a fix in the morning, but our only problem is that we have been unable to reproduce this on either sisi or local test servers. So looks like we are going to have to test that on TQ. but i have full faith in Papasmurf
Valar is looking into the current node death problem, not sure the details on it as there is nothing i can do to help him really and i've been doing other stuff. but it's being looked into.
Mike is now working on the memory leak problem for the market nodes and generally looking for leaks. We took some fresh market data from TQ which he is looking at and once the memory debugging is fixed, then he can watch the memory useage while it's live.
it's all go atm 
Spank You later |
|
| |
|