-- [ Picked text/plain from multipart/alternative ] On Windows we gave up and just made all our servers 18 & 20 player servers
On 4/24/07, Arne Guski <[EMAIL PROTECTED]> wrote: > > This is a multi-part message in MIME format. > -- > [ Picked text/plain from multipart/alternative ] > As it became quiet around this problem, i would also like to point out > that this issue is far from beeing resolved. > > I state again: > We run all-Intel-CPUs from P4s over dual / quad xeons to core 2 duos, > all with nearly the exact lag issues. > All running Debian almost the same problems on Sarge and Etch (now > running Etch since its stable) > All running cut down Preemt Kernels with 1000Hz (2.6.16.40+), also tried > countless other Versions including 2.6.20 and 2.6.21-rc7 with / without > dyntick, different IO Schedulers ... so everything that could cause > problems is sorted out from our side. > Since these boxes are dedicated for gameservers only there is not much > else on them besides proftpd-mysql, snmpd and our webinterface-backend > (really just some perl scripts croned) that could cause any trouble. > > Problems are almost identical with the current "public" version and the > beta, therefore our customer servers are still running on build 2949, as > everything else is performing far from what our customers expect. > > I agree with TC from UKCS, that valve should put some focus on > optimizing performance, but then ... they have not thought about that a > single second in the last 2 year .. so basically i have given up hope on > that one ... > > I do hope this issue will get resolved soon ... as there might be a > point where we will be unable to use the "working" build (caused by some > client update or whatnot) and therefore beeing unable to offer source > servers to our customers. > > Arne Guski > www.server-sales.de > Forum: http://forum.server-sales.de > Qnet: #server-sales.de > Mail: [EMAIL PROTECTED] > ICQ: 48395948 > > > > > > Message: 3 > > From: "Michael Shavit" <[EMAIL PROTECTED]> > > To: <[email protected]> > > Subject: RE: [hlds_linux] Still CSS Problems > > Date: Mon, 23 Apr 2007 20:22:49 +0200 > > Organization: Binarics GmbH > > Reply-To: [email protected] > > > > Hello, > > > > I would like to add here too from our side, we still have fps drops and > > latency increase even with the betaupdate, on Debian as also RedHat. > Hope > > Valve is still taking care of that. > > > > Cheers, > > Michael > > > > -----Original Message----- > > From: [EMAIL PROTECTED] > > [mailto:[EMAIL PROTECTED] On Behalf Of Tobias > Rehn > > Sent: Montag, 23. April 2007 13:19 > > To: [email protected] > > Subject: [hlds_linux] Still CSS Problems > > > > Hey, > > > > We tested several configurations (Debian unstable, testing, stable + > > different kernels + different hw [Intel <-> AMD]) the last weeks and > > there still seem to be these enormous fps drops in css. There is no > > difference between the beta binaries and the originals. Does anyone have > > a fix for this? > > > > I already tested debian unstable because of the new glibc (2.5). But > > there is still some loss of performance according to the server versions > > before the last update. It runs much better with this version but still > > its not as good as before the server update 1 1/2 weeks ago. > > > > Kind regards, Tobias Rehn > > > > _______________________________________________ > > To unsubscribe, edit your list preferences, or view the list archives, > > please visit: > > http://list.valvesoftware.com/mailman/listinfo/hlds_linux > > > > > > > > > > > > > > --__--__-- > > > > Message: 6 > > From: "TC" <[EMAIL PROTECTED]> > > To: <[email protected]> > > Cc: <[EMAIL PROTECTED]> > > Subject: RE: [hlds_linux] Still CSS Problems > > Date: Mon, 23 Apr 2007 21:39:07 +0100 > > Reply-To: [email protected] > > > > The srcds0407 source beta engine causes terrible lag problems at the > start > > of rounds on a full server and had to be backed off all my machines. It > > better not be included as a mandatory change in the next release as that > > would be disastrous. If Valve insists on that change they need to make > it > > optional controlled by a cvar, not forced on everyone. > > > > It is unclear to me why they started changing that part of the code to > > attempt a fix the problem in the beta, when it doesn't sound like that > was > > the area they had previously fiddled with. Valve shouldn't be changing > code > > to try to fix something in an area of the code they hadn't touched when > they > > broke it. That is only shifting the sands and in this case introduces > new > > problems for different people. > > > > I have to admit I have been fortunate not to have suffered from any of > the > > lag problems reported through this forum except from the beta srcds0407. > I > > run a mixture of operating systems on various boxes, both 32 bit and 64 > bit, > > AMD and Intel, single and multi processor and various kernel versions. I > use > > glib2.3.6 by default and also tried glib2.5 but have been unable to > > re-create the problem on any combination. > > > > The fact that it doesn't affect everybody makes it sound like they have > done > > something in the code that can now upset the operating system. If it was > > something basic it would be in the main valve code and affect everybody. > > Maybe they updated their compiler version or used different compile > > switches, or added a rogue system call with un-initialised parameters. > It's > > a bit like renice stopped being useable in a past valve engine release, > > that's probably due to the game upsetting the operating system and > something > > else they have not been able to track down. > > > > The only other thought on why UKCS has been unaffected maybe that we > only > > use our own custom mod not any off the shelf mods used on most other > > servers. Its won't be the mods themselves, but maybe a combination of > > libraries along with the latent bug introduced by Valve that causes it > to > > exhibit itself. However, I think some people say they could get the > problem > > without any mods but not sure if that was confirmed. Only Valve will > know > > ALL the changes they introduce between releases, and it's those areas > they > > changed including the build process they should review and only relating > to > > those areas they changed should they look to apply a fix. > > > > The biggest factor that affects the playability and popularity of the > > Valve's Source games is how smooth and tight their games feel to play. > I'd > > like to see more focus from Valve concentrating on improving performance > and > > underlying feel of the game as that's something that benefits all > players > > and what want above all other things enhancements. Too much time seems > to be > > spent making changes and introducing features that assume the basics are > > already in place and perfect, when they are not. Players want a game > they > > can play smoothly on the PC they already have at home which requires > making > > both the client and server side work with lower CPU and without hangs > and > > glitches. Continuing to develop the game in a direction that requires > ever > > more powerful Client PCs and graphic cards prices most of the potential > > players off the game. Similar goes for the amount of CPU the server side > now > > requires, now demanding the very top end available hardware to support > them > > which again out prices most and affects the end users experience who > play on > > them. I've ran Valve based game servers for five years, and CSS has > never > > ran as well as it did before the changes at the end of 2005. Back then > it > > ran with significantly less load on both client and server and felt > overall > > more enjoyable to play. > > > > Valve needs to concentrate on improving the fundamentals of their games, > and > > the number of players and popularity will go up which has to be Valve's > > reason for making changes. In contrast, I wonder how many more people > were > > drawn to play CSS by dynamic weapon pricing, HDR, those new text boxes > and > > some of those valve maps which were very pretty but too laggy for the > > average player? How many players benefited from the > > cl_restrict_server_commands constraint imposed on server mods against > how > > many were negatively affected by the knock on affects of introducing it? > > > > Did Valve ever fix the problem where the downloads directory of user > sprays > > gets very large over time and causes lag every time people spray while > it > > searches the directory to find the needed file? Or is each server owner > > still expected to know they need to provide their own regular purging? > I'd > > like to think Valve fixed that one, if not, why not. > > > > The better voice codec was on the right lines - focussing on the > > fundamentals of the game. Let's hope valve find their direction and > spend > > their coming development time on the top requirement of every player - > > improving the fundamental smoothness and feel of the core game itself. > > > > TC > > UKCS Mega Game Servers (www.ukcs.net) > > > -- > > _______________________________________________ > To unsubscribe, edit your list preferences, or view the list archives, > please visit: > http://list.valvesoftware.com/mailman/listinfo/hlds_linux > -- _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
