The bandwidth skyrockets because you're forcing clients to use bandwidth that the source client normally wouldn't use (I think the source defaults are something like 10 update/10 cmd/10000 rate, but I think they might just default to whatever the minimums of the server are unless client set, afaik.) Choke will always be there unless you've set the proper rates. The rates you're getting in is based on the amount of data the server is giving to you, for instance, if you're in the middle of a 8v8 player firefight, you're going to hit that 15000 maxrate in no time, causing you to get choke, while if you're by yourself in a corner you likely won't hit that limitation and thus won't get choke.
Basically choke is caused by the client not being able to receive the amount of updates that are required for your tickrate. If maxrate is set too low, clients will get choke, because they aren't getting the full amount of packets possible to ensure they're getting all player data at a reasonable rate. Bandwidth is going to skyrocket a bit... Remember, 10000 rate is ~10KB/s for each player at maximum (up/down.) Saint K. wrote: > HI, > > 1.) I adjust the client to the servers max limits to ensure proper testing > 2.) Servers available bandwidth is sufficient (100Mbps) > 3.) servers FPS seems in order to me (around 500fps, but fluctuating as > net_graph 5 shows) > 4.) Clients available bandwidth is good aswell (24Mbit/1.5Mbit) > 5.) A raw ping to the server returns under 8ms > > I have set your recommendations now, and indeed, with 50000 rate is seems to > be more or less good. One question tops my mind thought, why has the > bandwidth useage skyrocketed that much? > > And the other question which tops my mind is why the USA server seems to > have no choke issue's with just 2 players less as max, and a max rate > setting of 15000. Would the higher latency play a part in this? > > Cheers, > > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Cc2iscooL > Sent: Wednesday, July 16, 2008 7:11 PM > To: Half-Life dedicated Linux server mailing list > Subject: Re: [hlds_linux] Choke Problem > > Use net_graph 5 next time. This shows the server's current FPS as sv: > xxx in the lower left of the graph. > > Choke depends on a lot of things. > 1.) Client's rate settings. > 2.) Server's available bandwidth. > 3.) Server's FPS > 4.) Client's available bandwidth. > 5.) Quality of network/latency in-between client/server. > > My recommendations: > > I have 24 slot servers and have used these rates for a while, they work > great for most players and if the client sets their rates properly they > get no choke. > > //Server Side Rates > sv_maxrate 50000 //Yes, 50000. > sv_minrate 15000 //15000 is pretty modest for slower connections but > fast enough for no noticable lag. > sv_mincmdrate 20 //20 out of 66 (tickrate) should be enough for minimum. > sv_maxcmdrate 67 // Doesn't need to be more than 1+ tickrate. > sv_minupdaterate 20 //See mincmdrate > sv_maxupdaterate 67 //See maxcmdrate > > //Client Side Rates (Yes, you will have to set these.) > rate 50000 > cl_cmdrate 67 > cl_updaterate 67 > > I've found that I get choke with anything below 50,000 setting when > there are a lot of players on screen. With this setup I've yet to have > more than 1-2 choke in even the heaviest of firefights on my 24 slot > servers. > > There's another problem involved in this. If your server isn't holding > at least the tickrate in FPS (if your tickrate is 66 and your FPS > constantly dips to 20 on the server, it's not holding tickrate) then > you'll also get choke. Easy way to check this is while your server is > full (and players are active) type "stats" a few times into your console > (HLSW works great for this) and look at the FPS column. If your FPS > isn't holding 66 or better, you're going to get choke whether your rates > are set correctly or not. If this is the case, it's likely your CPU is > underpowered or the configured OS isn't configured properly for what > you're doing. > > Hope this helps, > Cc2iscooL > > Saint K. wrote: > >> Hi, >> >> We're currently experiencing a massive choke problem, and I can't seem to >> find the issue. >> >> Here is a screenshot of the net_grap 3 running; >> http://www.specialattack.net/downloads/choke.jpg >> >> All clients are experiencing these issues. >> >> We're having the choke issue's on both machines. One machine runs Debian >> Etch with a tickles kernel(2.2.23), and the other machine runs also Debian >> Etch, but with a 1000Hz/preemt kernel (2.2.26). >> >> Both the TF2 servers run around 500FPS. The CPU load on the core is around >> 60-70% tops. >> >> I've tried various rate settings, but the choke seems to be a near >> > constant, > >> even with a low amount of players. >> >> The server is connected on a 100Mbit uplink, which is without problems as >> far as we can see on the monitoring and other testing we performed. No >> latency spikes, packet loss etc etc. >> >> Does anyone have any tips were I should be looking to solve this choke >> problem? >> >> Cheers, >> >> >> _______________________________________________ >> 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 > > > No virus found in this incoming message. > Checked by AVG - http://www.avg.com > Version: 8.0.138 / Virus Database: 270.4.11/1553 - Release Date: 7/15/2008 > 5:48 AM > > > > > _______________________________________________ > 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
