I'm not really good with computers... How is usage relevant? To be it seems like you could have reduced performance with the same usage, but I don't really have a very in-depth understanding of this sort of thing. Do you mind explaining this? On Dec 9, 2015 6:51 PM, "Don Park" <[email protected]> wrote:
> Yikes. Seems I made a bit of a mistake there. > > 1. I forgot to mention, the dip on Wednesday near 00:00 was referencing > the memory usage graph. > > 2. Our average memory usage isn't that high. If you're familiar with > Linux, it allocates unused RAM as cache. If a program needs the RAM, linux > will then automatically free up the needed RAM for use. More reading can > be done here about this if you're interested: > http://www.linuxatemyram.com/ > > 3. Therefore, here's another graph of the exact same time period and > location from our other monitoring system: http://i.imgur.com/VU2K4Wd.png > . Again, the dip in the RAM usage is our CSGO server update script > running. > > > > > On Thu, Dec 10, 2015 at 8:42 AM, Don Park <[email protected]> wrote: > >> We're on Ubuntu 14.04, with the 3.13.0-63-generic kernel. >> >> I've attached some graphs and information from our monitoring system. >> All of them are the past week's monitoring (data collection time step of a >> minute). I believe the data collection time-zone is for Las Vegas, whereas >> this specific instance is located in Singapore. So please mind the time >> difference. As to note, our deployment doesn't put any limitations on any >> SRCDS game server instances, and our hardware is underloaded to make sure >> that our game server instances always have room to grow. This is a >> dedicated server after all, so we know 100% what's going on hardware-wise. >> >> Processor Usage (Week): http://i.imgur.com/oGqbnUv.png >> Memory Usage (Week): http://i.imgur.com/yTiFfFO.png >> Load Average (Week): http://i.imgur.com/Z9cDgTf.png >> >> To put the graphs and data into perspective, our current deployment is a >> Dual L5620 (so two quad core processors clocked at 2.4 GHz with >> hyperthreading comes to a total of 16 threads) node with 24 GB RAM. For >> the CSGO servers, we've made a KVM within the server with 8 CPU cores and 8 >> GB RAM. This specific KVM deployment runs around 6 CSGO Servers. >> >> You can see the dip on Wednesday near 00:00, that was the time I ran the >> update scripts. using the last week's historical data as the "baseline" >> point, you can see (on the Processor Usage graph) that the average CPU >> usage is still within historical max-used limits, however does seem to be >> on the higher end of the spectrum. I'd like to assume this is basically >> just more people who got back and played CSGO to check out the new update. >> >> If you look at the Load Average graph, you can see the same spike here, >> except larger (as load average is cumulative (or the total "sum" of the >> load on all 8 cores), whereas the processor usage is distributed (or as >> just shows the % usage of each core)). However, it's in line with Friday's >> numbers (Friday of course being one of the more popular days with higher >> user peaks). >> >> Processor Usage (Month): http://i.imgur.com/kjEsQPU.png >> Load Average (Month): http://i.imgur.com/JuezJVz.png >> >> Now here are the past month's numbers. There's nothing out of the >> ordinary that I can tell happening server hardware-wise. >> >> Now related to in-game statistics, I don't have the numbers on me but no >> major changes in variance (network-wise nor tick-wise) has been >> experienced. I myself when playing on our servers haven't seen any major >> differences or changes in the tick variance, nor have I received reports of >> the variance or anything else is out of the ordinary... besides for >> complaints that the R8 Revolver completely breaks the game. >> >> Let me know if there's anything else I can share with you. >> >> >> >> >> >> >> On Thu, Dec 10, 2015 at 8:07 AM, Absurd Minds <[email protected]> >> wrote: >> >>> I noticed it in the valve servers, which are also Ubuntu, but it's hard >>> to tell there because valve server performance is always so spotty. >>> On Dec 9, 2015 6:02 PM, "Absurd Minds" <[email protected]> wrote: >>> >>>> Are you also on Ubuntu? >>>> On Dec 9, 2015 5:48 PM, "Don Park" <[email protected]> wrote: >>>> >>>>> This is not the case for our deployment. >>>>> >>>>> >>>>> >>>>> On Thu, Dec 10, 2015 at 7:40 AM, Absurd Minds < >>>>> [email protected]> wrote: >>>>> >>>>>> I've noticed on my servers that the variance is spiking quite a lot >>>>>> since yesterday's update. I've noticed this in other servers, too (but >>>>>> obviously I have no way of knowing if that's abnormal to their server). >>>>>> >>>>>> Im running Ubuntu. >>>>>> >>>>>> _______________________________________________ >>>>>> Csgo_servers mailing list >>>>>> [email protected] >>>>>> https://list.valvesoftware.com/cgi-bin/mailman/listinfo/csgo_servers >>>>>> >>>>> >>>>> >>>>> _______________________________________________ >>>>> Csgo_servers mailing list >>>>> [email protected] >>>>> https://list.valvesoftware.com/cgi-bin/mailman/listinfo/csgo_servers >>>>> >>>> >>> _______________________________________________ >>> Csgo_servers mailing list >>> [email protected] >>> https://list.valvesoftware.com/cgi-bin/mailman/listinfo/csgo_servers >>> >> >> > > _______________________________________________ > Csgo_servers mailing list > [email protected] > https://list.valvesoftware.com/cgi-bin/mailman/listinfo/csgo_servers >
_______________________________________________ Csgo_servers mailing list [email protected] https://list.valvesoftware.com/cgi-bin/mailman/listinfo/csgo_servers
