| Pages: 1 [2] :: one page |
| Author |
Thread Statistics | Show CCP posts - 1 post(s) |
Bobby Atlas
Di-Tron Heavy Industries
Atlas Alliance
   |
Posted - 2010.01.06 11:31:00 -
[31]
Edited by: Bobby Atlas on 06/01/2010 11:31:58
<6am rant>
Let me start off with saying this was the most disorganized testing I have ever seen and I do not think anyone with even the slightest experience in Q/A or development took this seriously after the first 5 minutes.
- The participation on the test server was absolutely horrible - none of the tests even began to scale to what we are experiencing on tranq
- There was a constant stream of 30-50 people in the moveme channel, none of which were getting moved most of the time. I know more than a few people who got tired waiting and just gave up on the testing. In the future, dedicated individuals handling moves promptly would greatly increase participation.
- The effort required by players to actually get the pre-dominion client setup was, for most people, more effort than they were willing to commit - especially during the middle of the week.
- The actual scheduling time was also a big problem, for North Americans it was the middle of the work day and for our Euro buddies, the last thing they wanted to do after a long day of work was blow there evening on a half-arsed test. Btter scheduling such as conducting this over a weekend would have greatly helped with participation
- Throwing a all-skills-for-everyone style setup would draw far more players to participate, just give everyone max skills - do not seed any supercaps but obviously those whom already have them would bring them.
Then there is this thread, which seems to be more indicative of CCP grasping at straws either looking for a client side issue or outright putting this as a client side issue. I think the vast majority of people involved in the major 0.0 conflicts since dominion launched would all agree that the actual client in of itself is not an issue, it is from our perspective doing everything it is supposed to.
During the grid load issues, the client remains responsive, channels continue to operate, server moves client properly to new system, switches channels, no notable CPU or memory increases on grid loading or jumpins, actual FPS with eve has always been an unreliable and vastly inconsistent method of gauging performance.
Though some report jump black screens I think this is a minority of people, the same as those who claim "black screens on undock", perhaps insufficient hardware causing texture loading to be slow or poor performing system causing eve's in-memory cache to fetch data very slowly.
I think I can best describe this issue being server side as, it is not uncommon that you move a 200 man fleet and sometimes you will just hit a random system and BANG, that system CRAWLS and your entire fleet (just 200 of you trying to pass through the system gate to gate) are instantly stuck in there. We have had this (in atlas) happen more than a few times where fleets get bottle necked into a system for up to an hour with just trying to travel through it. Another example is an engagement of 800 people going on in pure blind or fountain and the nodes all the way in geminate become completely unworkable, from undocking to gate jumping to god forbid pvp.
If I was working at CCP I would be looking at the database layer as perhaps the most common interconnect between all the issues, the real sad reality of this whole affair is that we have seen little in the way of actual acknowledgment from CCP from someone who knows what they are talking about (no offense tanis). Where are all those pretty graphs you have available, show us the SQL load or per node CPU/Memory usage average or similar as compared pre and post dominion; surely CCP does not expect us to eat up this half arsed testing and "how is your FPS during the tests" as any serious effort to resolve these performance issues with the server side infrastructure.
</6am rant>
|
GoldenZah
|
Posted - 2010.01.06 11:41:00 -
[32]
- 5.1.2009
- Windows 7 Pro 64bit
- AMD Athlon 64 (3800+)
- Nvidia 9600GT (1GB ddr2)
- 2GB DDR
Experience
- 20fps
- Game sound: yes Voice: no
- Brackets off
- No - default setting
- 2
|
Salli Nerval
|
Posted - 2010.01.06 12:16:00 -
[33]
Edited by: Salli Nerval on 06/01/2010 12:16:23
100% aggree with bobby atlas, the trouble is not placed in the client. it is a server / db issue.
all this SiSi testing is nice entertainment but not sure if is it worth. TQ is a completly other cluster and server infrastructure.
Move this test to TQ, take eve "offline" one weekend. Make two parties and give them goals and rewards. so you have enough members and real conditions. after that rollback the server 2 days. i am sure non of the 0.0 pilots will be sad about. |
Batolemaeus
Caldari
Free-Space-Ranger
Morsus Mihi
|
Posted - 2010.01.06 16:04:00 -
[34]
Edited by: Batolemaeus on 06/01/2010 16:04:15
march 5th, only on sisi, not multi
Win2k8 64 bit
Amd Ahtlon 64 X2 6000+ @ 3ghz
Gf8600gt
4gb ram
Average fps between 1 and 30. Whoever thought it was a great idea to set "show all brackets" to on as default needs a reality check. Took me a while to find it and switch it off. Unplayable without very aggressive bracket settings disabling almost anything.
No Audio
Interval One on both clients.
Can't really comment on performance. Played like standard roaming gang combat in lowsec..
Comments:
Not enough lag was produced. I never experienced stuck guns, grid not loading or ghost ships. Compared to the lagstorm on tq after dominion, this lag was just a slight breeze.
Also, why is "broadcast to practically nobody" a standard setting for broadcast range? It made logistics much harder. Default bracket settings suck too.
----------------------------------------------
 Originally by: CCP Prism X
In New Eden, EVE wins you.
|
Caladain Barton
Navy of Xoc
Wildly Inappropriate.
|
Posted - 2010.01.07 03:11:00 -
[35]
Originally by: Salli Nerval
Edited by: Salli Nerval on 06/01/2010 12:16:23
100% aggree with bobby atlas, the trouble is not placed in the client. it is a server / db issue.
all this SiSi testing is nice entertainment but not sure if is it worth. TQ is a completly other cluster and server infrastructure.
Move this test to TQ, take eve "offline" one weekend. Make two parties and give them goals and rewards. so you have enough members and real conditions. after that rollback the server 2 days. i am sure non of the 0.0 pilots will be sad about.
I too 100% agree with Bobby..and with your statement. Contact Coalition leaders, and do this *right* with a bit of thought..not haphazard..
Hell, any 0.0 fleet commander can tell you exactly how to recreate the situation *on demand*...this is not rocket science..we've figured it out by losing fleets to the evil grid lag, and learning from others who have lost fleets to evil grid lag.
|
Mani Fug
Majesta Empire
|
Posted - 2010.01.07 18:12:00 -
[36]
5th January 2010
WIN XP Professional
AMD X2 6100+
GPU NVidia 9800 GTX
2GB DDR2 800
òDid you have Audio enabled?
Yes, EvE Voice, too
òDid you have brackets turned on?
No
Both Tests with same Settings.
Apocrypha-Code:
Small Module Lag, but was really well playable
Dominion-Code:
After the destruction of my BS I could not do anything. POD is not allowed to move. Same BS exploded several several times. Just the chat channels are still working. For this reason, I could not really participate in the second Test. After restarting the game I could easily bridge up to the fleet. All modules have worked.
One would have to announce this test on the homepage, so were able to attend really many players. The test itself was a bit chaotic.
|
SupaKudoRio
|
Posted - 2010.01.08 07:46:00 -
[37]
Excuse me while I try to bend my mind around why mass testing is even needed; other MMO developers use automated sequences (i.e. having a test server run a simulation of a massive fleet battle using simple AI, which interfaces with the server through the normal client protocols from a separate machine, and a handful of mock clients performing whatever needs testing via a bot.
Running a test like that for a few hours will get you enough data to find and fix a redonkulous heap of bugs.
Or is what everyone else is capable of doing so utterly beyond your ability that automated testing wasn't even considered...?
/rant
I didn't even know testing was being done. 
---------------
10/10: Where is your God now? |
Ballistic56
|
Posted - 2010.01.09 03:51:00 -
[38]
I took part at the second test:
Client setup doesn't really matter as it was all smooth.
There were too few players as most people have pointed out.
After the many NPCs were spawned there was considerable lag, i was the next ship to get blown up at that time. Lag was about 30 seconds i guess.
After hitting "undock" i had to wait a few minutes with the "black screen" (intersting though: there was a mail notification while the rest of the screen was black).
It might not be possible to get more than about 200 people without doing this at the weekend or with a special reward or something, so i suggest for the next lag testing event:
get that test server software on the worst hardware you can find! if it can barely handle 100 people in system it will be perfect to try to fight 100 vs 100 with all those really bad consequences many people experience in really big fleet fights.
|
MAX MEXX
|
Posted - 2010.01.19 10:46:00 -
[39]
What about having a TQ test?
(or just check 49u- and the systems in the north)
|
Carniflex
StarHunt
Systematic-Chaos
|
Posted - 2010.01.19 12:28:00 -
[40]
Originally by: MAX MEXX
What about having a TQ test?
(or just check 49u- and the systems in the north)
There is almost every bigger battle in there a Polaris frigate sitting. Observing. And propably getting as bad lag. I sure hope they get some good data our from there. Hell, with us sitting in dedicated node for the past week they even could have been trying different tweaks on the server side to see what happens.
Have not seen em down in the HED area tho, but they do have rather fancy cloaks  .
|
| |
|
| Pages: 1 [2] :: one page |
| First page | Previous page | Next page | Last page |