> -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf > Of [EMAIL PROTECTED] > Sent: 28 November 2008 10:25 > To: 'Half-Life dedicated Linux server mailing list' > Subject: Re: [hlds_linux] -fork configuration tip (dominic) > > ...and the list software appears to have stripped my > attachments... Will try again > > > -----Original Message----- > > From: [EMAIL PROTECTED] > > [mailto:[EMAIL PROTECTED] On Behalf Of > > [EMAIL PROTECTED] > > Sent: 28 November 2008 10:22 > > To: 'Half-Life dedicated Linux server mailing list' > > Subject: Re: [hlds_linux] -fork configuration tip (dominic) > > > > These are probably the two most relevant emails... (The contents of > > the first - Chris Green's - was in a readme file with the > server, but > > that appears to have been in the demo install only). The second - > > Milton Ngan's > > - is about managing large amounts of servers using fork, and is > > slightly outdated (matchmakingport and systemlinkport don't > exist any > > more). If you use fork and netconport, there are a few command you > > can run from the parent console - use "find" to get a list of these. > > > > Hope this is of use > > > > > -----Original Message----- > > > From: [EMAIL PROTECTED] > > > [mailto:[EMAIL PROTECTED] On > > Behalf Of Nander > > > Paardekooper > > > Sent: 28 November 2008 10:07 > > > To: Half-Life dedicated Linux server mailing list > > > Subject: Re: [hlds_linux] -fork configuration tip (dominic) > > > > > > So if i start a server with -fork 1 and one with -fork 2 > i see only > > > one process or...? HOWTO! ;-). > > > > > > Best regards, > > > > > > Nander Paardekooper. > > > > > > Ferenc Kovacs schreef: > > > > On Thu, Nov 27, 2008 at 5:00 PM, Nander Paardekooper > > > > <[EMAIL PROTECTED]>wrote: > > > > > > > >> And now I have a question ;-) > > > >> > > > >> What the hell can i do with "-fork", because i cannot find any > > > >> information about it on any hlds site or manual. > > > >> > > > >> Best regards, > > > >> > > > >> Nander Paardekooper. > > > >> > > > >> J T schreef: > > > >>> You should have replied to the original thread, easier > > for people > > > >>> who go through the list archives to find valuable > > > information like this. > > > >>> > > > >>> On Thu, Nov 27, 2008 at 6:57 AM, Dominic Baranski < > > > >>> [EMAIL PROTECTED]> wrote: > > > >>> > > > >>>> I was waiting for someone to actually talk about how > > > load average > > > >>>> works.. Thank you! You're the first person to > > actually mention > > > >>>> that load average is the "length of the process queue on the > > > >>>> system". Now, if anyone looks into this further they will > > > >>>> immediately find that load average for one system will > > > not mean the same for another.. > > > >>>> > > > >>>> For example > > > >>>> http://s02.legionofboom.org/monitorix/imgs/cpu1.week.png > > > >>>> shows my system at LA's hitting well over 10 (during > > busy times) > > > >>>> and I nor anyone else on my servers have experienced > > > lag. (thats > > > >>>> running with a -fork 20 btw) > > > >>>> > > > >>>> So my point is.. load average is only a possible > indicator of > > > >>>> problems and should not be taken as the "end all" for > > > performance monitoring. > > > >>>> > > > >>>> On Wed, 2008-11-26 at 10:50 -0800, > > > >>>> [EMAIL PROTECTED] wrote: > > > >>>>> Message: 1 > > > >>>>> Date: Wed, 26 Nov 2008 12:15:46 -0600 > > > >>>>> From: [EMAIL PROTECTED] > > > >>>>> Subject: Re: [hlds_linux] hlds_linux Digest, Vol 9, > Issue 199 > > > >>>>> To: [email protected] > > > >>>>> Message-ID: > > > >>>>> > > > >>>>> > <[EMAIL PROTECTED]> > > > >>>>> Content-Type: text/plain; charset=ISO-8859-1 > > > >>>>> > > > >>>>> Keep in mind that load average is an expression of > > > queue depth. > > > >>>>> Some types of servers can handle deep queues with > no visible > > > >>>>> impact shown to the end user. Gameservers are not > > that kind of > > > >>>>> animal. They require constant, nearly instantaneous > > updates in > > > >>>>> both directions in order to have truly smooth, seamless > > > play. In > > > >>>>> the past, any time I saw load average above one I saw > > > an impact of > > > >>>>> some sort. > > > >>>>> > > > >>>>> A load average above 3 seems to manifest itself in L4D > > > servers in > > > >>>>> some unpredictable ways that can generally be ignored > > > by most players. > > > >>>>> > > > >>>>> If you don't want to keep sending jarring notes that > > throw your > > > >>>>> players off, you really want to see your load > average below 2. > > > >>>>> This should be true regardless of the number of CPUs > > > that you have > > > >>>>> in your box. Thus, my comment in an earlier email about > > > >>>>> converting one of my public servers back over to Steam > > > group only. > > > >>>>> Having 2 dedicated to just the Steam group will still > > > let people > > > >>>>> fill up the public servers first and use the two > > > semi-private ones > > > >>>>> as relief. > > > >>>>> > > > >>>>> > > > >>>>> [snip] > > > >>>>> > > > >>>> _______________________________________________ > > > >>>> 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 > > > >> > > > > the forked process has access for his parents memory > > > scope.so you can > > > > decrease the running instances memory footprint this way. > > > > > > > > Tyrael > > > > _______________________________________________ > > > > 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 > > > > > > > > _______________________________________________ > To unsubscribe, edit your list preferences, or view the list > archives, please visit: > http://list.valvesoftware.com/mailman/listinfo/hlds_linux >
From: Milton Ngan <[EMAIL PROTECTED]> To: "[email protected]" <[email protected]> Date: Tue, 11 Nov 2008 15:58:16 -0800 Subject: [hlds_linux] l4d Dedicated server notes
I have been working with Chris to get all of the dedicated servers up and running with the new features. What I have been using to host dedicated servers for L4D is: 2x 2.66 GHz Core 2 Quad (Yorkfield) or 2x 2.5GHz LV Core 2 Quad (Yorkfield) 8GB RAM Debian Lenny (2.6.26-1-amd64) On this configuration I have been able to run 40 instances of the L4D dedicated server supporting 120-140 players simultaneously. With Vs mode, this number will increase with no noticeable change in performance. The typical command line I am using to start up the servers in -fork mode looks like this: ./srcds_run \ -game left4dead \ -ip 1.1.1.1\ -netconport 9000+##\ -netconpassword password\ -steamport 27690+##\ +hostname My_L4D_Server_##\ +sv_tags reservable\ +exec server/##.cfg\ +map l4d_dem_hospital01_apartment\ -fork 40 I think what Chris forgot to mention was that the ## syntax can be used in an expression. So here rather than just substituting the instance number in, I am using it to add the value to an existing number. Someone asked about why -exec ##.cfg didn't work. I would suggest using +exec ##.cfg instead. This is the format I am using and this appears to function correctly. Also, make sure that your files are zero padded to two digits (i.e 01, 02, etc...). I also had to do a lot of shuffling around of port numbers to allow 40 server instances to run correctly. This is because many of the default port numbers are too close to one another to support more than about 5-10 servers. The ports you need to worry about are: -netconport -steamport +clientport +hostport +tv_port +matchmakingport +systemlinkport I control the first two from the command line since they are not convars, the rest I put into my config files. Just space the port nunbers apart so there is a gap larger than the number of dedicated servers you intend to run on the same box. Milton _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
From: Chris Green <[EMAIL PROTECTED]> To: "'[email protected]'" <[email protected]> Date: Wed, 5 Nov 2008 17:36:35 -0800 Subject: [hlds_linux] l4d Dedicated server notes Linux dedicated server changes for Left4Dead There have been many changes, enhancements, and optimizations with the dedicated server for Linux: * All of the code is now being compiled with gcc 4.3.0 and glibc 2.8-8 in order to take advantage of compiler fixes and optimization enhancements. In order to run the l4d Linux dedicated server, you will need a system which can run binaries built against this version of glibc. * As a performance enhancement, the dedicated server can now use the hardware "RDTSC" timer instead of gettimeofday() for its internal timing, on systems which support it ( such systems are identified by having the "constant_tsc" flag set in /proc/cpuinfo ). When the code detects that the system has this, it will execute a benchmark to measure the actual rate of the hardware timer, and use this for all timing. If this causes trouble on a system, you can set the environment variable "RDTSC_FREQUENCY" to "disabled". * The Linux dedicated server is now capable of running multiple server instances as sub-processes off of one parent process. This provides a memory savings (through sharing of read-only data), a speedup when starting multiple servers, and also enhances server stability by having the servers restart as new sub-processes after each game is completed. This is controlled by the "-fork n" option. For instance, you can run "srcds_run -fork 5" to start up 5 separate server instances in this mode. * When -fork mode is enabled, some options on the command line can be parametrized based upon the server instance, by typing '##' on the launching line. For instance, "-fork 10 -netconport 90##" would cause the first server instance to use port 9001, the second to use 9002, etc. * There is a new remote access feature which allows a server administrator (or tool) to create a persistent connection to a running server, and to see its entire console output and send it commands. This feature is enabled by using the "-netconport" option when launching srcds_i486. For instance, if a server is started with "-netconport 9000", someone with access to port 9000 of the server could type "telnet xxxx 9000" and view the console output. This feature is disabled by default, and should only be used when an appropriate firewall/tunnel is used to control access to this port. Multiple clients can connect to the netconport of a running server at the same time. * If the netconsole is enabled, and the option "-netconpassword xxx" is set, the network console will not execute commands received through the netconport until the command "PASS xxx" is entered. _______________________________________________ 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
