Re: [hlds_linux] 1.6 hitbox issues
I would also be looking at the scheduler and OS. With that amount of "timecritical processes" you need a OS with a sched that are up to the task along side with IO. (not just disk IO) Load is not that good of a messurement for this. There should be other tools that shows how well it is working. Load should more just be considered as an indicator when it goes haywire. There are a reasons why you just run one service per server in the "real world". I would not even consider to run 24 csservers on a single server. The more csservers you put in, the more stress it will be for the OS. And since you only uses aprox 25% of all slots. go down to 8 servers and move your way up until you find when the issues occur. Then find the reason why it occured. Setting up AMXX ontop of that will lead to even more workload for the OS, especially if you fork something or uses it to connect and process data to and from a db. Might be worth mentioning that you might need to look at the network configs of the OS. I constantly upgrade my servers whenever I find (or think I have) an improvement to the scheduler. But then again, I am on FreeBSD. Once 7.0 is released, I will switch all my servers to that OS and the new scheduler. /Bjorn On Wed, 17 Oct 2007, Regime wrote: > With that many separate servers running, I'd have a look at the disk IO > on the server. The package iostat might come in handy for that. > (provided through yum in the sysstat package on RH based OS's.) > Gr, Regime > > Jonathan Selander wrote: > > load average: 2.10, 1.59, 1.39 > > > > Since i've got 8 cores it doesn't seem right to worry about 2.10 load, > > does it? > > > > It's running 24 servers and a maximum of 264 slots (which has never been > > reached, the average right now is around 60 slots used) > > ___ > 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
Re: [hlds_linux] 1.6 hitbox issues
With that many separate servers running, I'd have a look at the disk IO on the server. The package iostat might come in handy for that. (provided through yum in the sysstat package on RH based OS's.) Gr, Regime Jonathan Selander wrote: load average: 2.10, 1.59, 1.39 Since i've got 8 cores it doesn't seem right to worry about 2.10 load, does it? It's running 24 servers and a maximum of 264 slots (which has never been reached, the average right now is around 60 slots used) ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] 1.6 hitbox issues
load average: 2.10, 1.59, 1.39 Since i've got 8 cores it doesn't seem right to worry about 2.10 load, does it? It's running 24 servers and a maximum of 264 slots (which has never been reached, the average right now is around 60 slots used) Steve S. skrev: How does your CPU usage look? How many servers / slots are you running on your box? It's a dell 1950 machine with broadcom gbit cards, and it has 8gb memory. The servers run a couple of amxx mods, can this be the reason? That hlds is doing too much stuff in the background or so. rowan skrev: Jonathan Selander wrote: I run a bunch of CS 1.6 servers on a dual quadcore xeon machine. Lately my customers have been complaining about hitbox issues. What can be the cause of this? ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux numours causes: overloading server. that machine should have a gb network card. if its linux make sure ur not running hltv, iirc there is still a memory leak issue. ___ 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 ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] 1.6 hitbox issues
How does your CPU usage look? How many servers / slots are you running on your box? > It's a dell 1950 machine with broadcom gbit cards, and it has 8gb > memory. The servers run a couple of amxx mods, can this be the reason? > That hlds is doing too much stuff in the background or so. > > rowan skrev: >> Jonathan Selander wrote: >>> I run a bunch of CS 1.6 servers on a dual quadcore xeon machine. Lately >>> my customers have been complaining about hitbox issues. What can be the >>> cause of this? >>> >>> ___ >>> To unsubscribe, edit your list preferences, or view the list archives, >>> please visit: >>> http://list.valvesoftware.com/mailman/listinfo/hlds_linux >> numours causes: overloading server. that machine should have a gb >> network card. >> >> if its linux make sure ur not running hltv, iirc there is still a memory >> leak issue. >> >> >> ___ >> 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
Re: [hlds_linux] 1.6 hitbox issues
It's a dell 1950 machine with broadcom gbit cards, and it has 8gb memory. The servers run a couple of amxx mods, can this be the reason? That hlds is doing too much stuff in the background or so. rowan skrev: Jonathan Selander wrote: I run a bunch of CS 1.6 servers on a dual quadcore xeon machine. Lately my customers have been complaining about hitbox issues. What can be the cause of this? ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux numours causes: overloading server. that machine should have a gb network card. if its linux make sure ur not running hltv, iirc there is still a memory leak issue. ___ 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
Re: [hlds_linux] 1.6 hitbox issues
Jonathan Selander wrote: I run a bunch of CS 1.6 servers on a dual quadcore xeon machine. Lately my customers have been complaining about hitbox issues. What can be the cause of this? ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux numours causes: overloading server. that machine should have a gb network card. if its linux make sure ur not running hltv, iirc there is still a memory leak issue. ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
[hlds_linux] 1.6 hitbox issues
I run a bunch of CS 1.6 servers on a dual quadcore xeon machine. Lately my customers have been complaining about hitbox issues. What can be the cause of this? ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux