-- [ Picked text/plain from multipart/alternative ] sv_minupdaterate "0" didn't help
here's another log of "rcon stats" taken in 2 second intervals after round start on a map called surf_bigramp_v2b with 23 players: http://www.mahalos.net/surf_bigramp_v2b.txt what's strange is that the CPU usage increases when the FPS jumps from 10 to 32 On 12/7/05, Whisper <[EMAIL PROTECTED]> wrote: > > -- > [ Picked text/plain from multipart/alternative ] > Of course we are > > But you can run rcon stats pretty fast, but in any case as the rcon stats > line you have posted here illustrates the problem. > > The CPU is just above average, but not abnormally high and not what you > would expect to see if your server fps crashes. > > Oh well, looks like we are all in the same boat and have to wait for Valve > to come out with a fix. > > Hmmm, and if I read your notes correctly then the sort term fix is to set > the sv_minupdaterate back to default i.e. 0. > > I'll give it a try. > > I threw the map thing in as I thought it might have something to do with > de_nuke or the running of de_nuke on the same box causing the all SRCDS > processes across the server to lag, without actually showing anything in > CPU > stats. > > Also the lag lasts a lot longer than the resolution of most of the > monitoring tools we use, i.e. Several Seconds. > > Its still annoying as hell, for obvious reasons > > Thanks. > > > > On 12/8/05, James Tucker <[EMAIL PROTECTED]> wrote: > > > > 57.36 40667.01 105291.80 118 3 2.04 20 > > > > Ouch, I've not seen a <10 reading there before. > > > > You have got your kernel timer resolution setup properly? > > (Running windows media player / flash / boot.ini switch / one of the > > windows "pingbooster" tools?) > > > > > > > > Yup wrote: > > > > >-- > > >[ Picked text/plain from multipart/alternative ] > > >i have been dealing with this same issue since September or October; > low > > FPS > > >even though CPU usage is low. it's the worst at round start when more > > than > > >20 players are playing, but after a few players die the FPS climbs back > > to > > >normal. > > > > > >machine specs: dual Xeon 2.8GHz, 2GB RAM, Windows 2000 > > > > > >SRCDS specs: CS: Source, 24 players, 66 tickrate, set at "AboveNormal" > > > > > >server.cfg: sv_maxrate "20000", sv_minrate "2000", sv_maxupdaterate > "66", > > >sv_minupdaterate "10", fps_max "300" > > > > > >- occurs on all maps, but it's worse on surf maps and custom maps with > a > > lot > > >of objects (like breakable glass) > > >- according to rcon stats, CPU load is about 12% with 1.5GB of free RAM > > at > > >round start when the FPS is at 10 > > >- Hyper Threading is OFF > > >- 100 Mbps up/down link > > >- same problem running at 33 tickrate > > >- setting SRCDS to "High" doesn't help > > >- same problem without any plugins running > > >- banned_user.cfg is 0 KB in size > > > > > >here's a log of "rcon stats" taken over a period of about 25 seconds > > after > > >round start while playing a map called breakfloor with 20 players: > > >http://www.mahalos.net/breakfloor.txt > > > > > >On 12/7/05, Whisper <[EMAIL PROTECTED]> wrote: > > > > > > > > >>-- > > >>[ Picked text/plain from multipart/alternative ] > > >>This is probably going to sound crazy, but it appears some of you have > > >>probably experienced this already. > > >> > > >>It is the type of lag that is typically associated with an overloaded > > CPU, > > >>where a player keeps getting dragged backwards. > > >> > > >>The net_graph 3 on the clients shows that there is little choke, > meaning > > >>low > > >>single digits at worse, no loss, and the IN/OUT KB/s is reasonably > > normal, > > >>only the IN Updates per second drop from the normally steady 66 > Updates > > >>per > > >>second (It is a 66 tickrate server) down to 50 or 40 odd per second, > but > > >>still nothing that usually is as normal indication of overloaded box. > > >> > > >>The Windows server CPU stats does not show a huge CPU spike at or > around > > >>the > > >>same time. > > >> > > >>Moreover, an rcon stats of the box at the time the lag is occuring, > > shows > > >>the CPU utilisation of around 50%- 60%, not the usual 85%-95% which is > > >>what > > >>you normally see when this type of lag hits a server when its > > overloaded. > > >> > > >>The only odd thing is that the server fps as stated by the rcon stats > > >>command at the time the lag occurs has the server only running at > around > > >>10fps. > > >> > > >>Obviously 10fps is bad, but none of the other usual suspects that > > normally > > >>indicate an overloaded box show up as being outside normal parameters. > > >> > > >>Also 10fps usually doesn't cause the drag back type of lag, which is > as > > >>most > > >>of you familiar with running SRCDS is a pretty fair indication of an > > >>overloaded CPU, which would be fine, except other than the fps > dropping > > >>like > > >>a rock, no other measurements or benchmarks you would normally use, > > >>indicate > > >>there is in fact a problem. > > >> > > >>How many of you are getting this sort of lag? > > >> > > >>What map(s) are you running at the time it occurs? > > >> > > >>Some body mentioned on the STEAM Forums that it only takes 70% CPU as > > >>measured by Windows for the 10fps bug to occur. > > >> > > >>BTW, at this moment we are running zero plugins for obvious reasons. > > >> > > >>Thanks > > >>-- > > >> > > >>_______________________________________________ > > >>To unsubscribe, edit your list preferences, or view the list archives, > > >>please visit: > > >>http://list.valvesoftware.com/mailman/listinfo/hlds > > >> > > >> > > >> > > >-- > > > > > >_______________________________________________ > > >To unsubscribe, edit your list preferences, or view the list archives, > > please visit: > > >http://list.valvesoftware.com/mailman/listinfo/hlds > > > > > > > > > > > > > > > _______________________________________________ > > To unsubscribe, edit your list preferences, or view the list archives, > > please visit: > > http://list.valvesoftware.com/mailman/listinfo/hlds > > > -- > > _______________________________________________ > To unsubscribe, edit your list preferences, or view the list archives, > please visit: > http://list.valvesoftware.com/mailman/listinfo/hlds > -- _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds
