VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
-
Edge100x
- Founder
- Posts: 13109
- https://www.youtube.com/channel/UC40BgXanDqOYoVCYFDSTfHA
- Joined: Thu Apr 18, 2002 11:04 pm
- Location: Seattle
- Contact:
Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
Python, I've also been testing a new version of Xen in the background, and it contains some additional optimizations. So far, running a test NS2 server, it offers promising performance gains. If the testing continues to go well and there aren't any show-stoppers, we could schedule a maintenance to apply it to the physical machine hosting your VDS.
Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
Thanks Edge.
If all goes well tomorrow morning I will rebuild the serve / install 12.04 and see how results go. If results are still the same i'd be happy to test the new xen version.
If all goes well tomorrow morning I will rebuild the serve / install 12.04 and see how results go. If results are still the same i'd be happy to test the new xen version.
Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
Server is being rebuilt as i type this.
Results will follow shortly.
Results will follow shortly.
Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
Results look to appear about the same. When i head home for lunch in 1.5hrs i'll hop on the server and see how net_graph 4 is representing the server fps.
Here is what i have so far (again completely fresh install)
And here is the server running pl_goldrush with 32 players again:
Here is what i have so far (again completely fresh install)
Code: Select all
xxx@yyy:~# lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 12.04.1 LTS
Release: 12.04
Codename: precise
xxx@yyy:~# uname -a
Linux railbait 3.2.0-24-generic #39-Ubuntu SMP Mon May 21 16:52:17 UTC 2012 x86_64 x86_64 x86_64 GNU/Linux
xxx@yyy:~# dmesg | egrep -i 'xen|front'
[ 0.000000] DMI: Xen HVM domU, BIOS 4.1.3-rc2-pre 06/08/2012
[ 0.000000] Hypervisor detected: Xen HVM
[ 0.000000] Xen version 4.1.
[ 0.000000] Xen Platform PCI: I/O protocol version 1
[ 0.000000] Netfront and the Xen platform PCI driver have been compiled for this kernel: unplug emulated NICs.
[ 0.000000] Blkfront and the Xen platform PCI driver have been compiled for this kernel: unplug emulated disks.
[ 0.000000] ACPI: RSDP 00000000000ea020 00024 (v02 Xen)
[ 0.000000] ACPI: XSDT 00000000fc00eaf0 0003C (v01 Xen HVM 00000000 HVML 00000000)
[ 0.000000] ACPI: FACP 00000000fc00e8d0 000F4 (v04 Xen HVM 00000000 HVML 00000000)
[ 0.000000] ACPI: DSDT 00000000fc002c40 0BC09 (v02 Xen HVM 00000000 INTL 20090123)
[ 0.000000] ACPI: APIC 00000000fc00e9d0 000D8 (v02 Xen HVM 00000000 HVML 00000000)
[ 0.000000] ACPI: HPET 00000000fc00eab0 00038 (v01 Xen HVM 00000000 HVML 00000000)
[ 0.000000] Booting paravirtualized kernel on Xen HVM
[ 0.000000] Xen HVM callback vector for event delivery is enabled
[ 0.080345] Xen: using vcpuop timer interface
[ 0.080354] installing Xen timer for CPU 0
[ 0.172070] installing Xen timer for CPU 1
[ 0.264073] installing Xen timer for CPU 2
[ 0.479939] xen/balloon: Initialising balloon driver.
[ 0.479955] xen-balloon: Initialising balloon driver.
[ 0.496281] Switching to clocksource xen
[ 0.501204] xen: --> pirq=16 -> irq=8 (gsi=8)
[ 0.501253] xen: --> pirq=17 -> irq=12 (gsi=12)
[ 0.501281] xen: --> pirq=18 -> irq=1 (gsi=1)
[ 0.501312] xen: --> pirq=19 -> irq=6 (gsi=6)
[ 0.501347] xen: --> pirq=20 -> irq=4 (gsi=4)
[ 0.501383] xen: --> pirq=21 -> irq=7 (gsi=7)
[ 0.529147] xen: --> pirq=22 -> irq=23 (gsi=23)
[ 0.574202] xen: --> pirq=23 -> irq=28 (gsi=28)
[ 0.574205] xen-platform-pci 0000:00:03.0: PCI INT A -> GSI 28 (level, low) -> IRQ 28
[ 0.893716] blkfront device/vbd/768 num-ring-pages 1 nr_ents 32.
[ 0.895781] vbd vbd-5632: 19 xenbus_dev_probe on device/vbd/5632
[ 0.897781] vbd vbd-5696: 19 xenbus_dev_probe on device/vbd/5696
[ 0.899167] blkfront: xvda: barrier: enabled
[ 0.909305] Initialising Xen virtual ethernet driver.
[ 0.913253] xen_map_pirq_gsi: returning irq 23 for gsi 23
[ 0.914061] xen: --> pirq=24 -> irq=23 (gsi=23)
[ 0.946850] XENBUS: Device with no driver: device/vfb/0
[ 0.947617] XENBUS: Device with no driver: device/vbd/5632
[ 0.948359] XENBUS: Device with no driver: device/vbd/5696
Code: Select all
10:38:50 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 128.96 843.68 123 5 55.91 32 132
10:38:50 stats
10:38:50 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 128.61 850.73 123 5 67.91 32 132
10:38:50 stats
10:38:50 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 129.92 866.56 123 5 45.60 32 132
10:38:51 stats
10:38:51 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 130.51 879.76 123 5 48.56 32 132
10:38:51 stats
10:38:51 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 130.53 888.29 123 5 65.57 32 132
10:38:51 stats
10:38:51 stats
10:38:51 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 130.69 897.16 123 5 52.48 32 132
10:38:51 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 129.87 905.24 123 5 57.11 32 132
10:38:51 stats
10:38:51 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 129.71 909.94 123 5 60.20 32 132
10:38:51 stats
10:38:52 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 129.07 921.61 123 5 55.50 32 132
10:38:52 stats
10:38:52 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 129.02 933.26 123 5 53.86 32 132
10:38:52 stats
10:38:52 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 129.10 948.35 123 5 63.12 32 132
10:38:52 stats
10:38:52 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 130.03 953.67 123 5 46.48 32 132
10:38:52 stats
10:38:52 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 131.16 959.51 123 5 55.48 32 132
10:38:53 stats
10:38:53 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 132.46 957.24 123 5 16.10 32 132
10:38:53 stats
10:38:53 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 132.69 946.02 123 5 68.45 32 132
10:38:53 stats
10:38:53 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 132.72 937.49 123 5 40.84 32 132 Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
Picture of HTOP:
^^again this is still on pl_goldrush.
(Note: This has same results on other maps i'm just using goldrush since i compared results yesterday on this map)
^^again this is still on pl_goldrush.
(Note: This has same results on other maps i'm just using goldrush since i compared results yesterday on this map)
Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
Have you tried sv_parallel_snapshot 1?
Don't mean to sound like a broken record, but it was a huge difference for us.
Don't mean to sound like a broken record, but it was a huge difference for us.
Not a NFO employee
Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
Soja,
I will try that right now - once the server changes map it will take effect.
Again, i'll post results once sv_parallel_packentities 1 is in place.
I will try that right now - once the server changes map it will take effect.
Again, i'll post results once sv_parallel_packentities 1 is in place.
Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
Sorry, I meant,
sv_parallel_sendsnapshot 1
sv_parallel_sendsnapshot 1
Not a NFO employee
Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
Whoops i pasted the wrong line. Current configuration:
And sv_fps playing cp_gravelpit:
I'll set goldrush as nextmap and get results again on that map.
Code: Select all
net_queued_packet_thread 1
sv_parallel_packentities 1
sv_parallel_sendsnapshot 1
net_splitpacket_maxrate 50000
net_maxcleartime 0.01
sv_max_queries_sec_global 100000
And sv_fps playing cp_gravelpit:
Code: Select all
13:19:18 stats
13:19:18 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 145.27 832.83 284 12 53.84 32 304
13:19:19 stats
13:19:19 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 141.68 845.10 284 12 62.65 32 304
13:19:20 stats
13:19:20 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 142.46 846.44 284 12 57.58 32 304
13:19:20 stats
13:19:20 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 143.07 860.92 284 12 44.92 32 304
13:19:20 stats
13:19:21 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 143.42 865.40 284 12 37.68 32 304
13:19:21 stats
13:19:22 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 143.03 856.17 284 12 44.65 32 304
13:19:22 stats
13:19:22 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 142.30 861.93 284 12 44.07 32 304
13:19:23 stats
13:19:23 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 142.14 872.72 284 12 63.92 32 304
13:19:23 stats
13:19:23 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 142.58 879.37 284 12 46.90 32 304
13:19:23 stats
13:19:24 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 142.98 899.71 284 12 45.47 32 304
13:19:24 stats
13:19:24 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 143.07 916.21 284 12 69.25 32 304
13:19:24 stats
13:19:24 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 141.70 943.31 284 12 56.45 32 304
13:19:25 stats
13:19:25 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 140.70 950.76 284 12 64.91 32 304
13:19:25 stats
13:19:26 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 138.85 928.67 284 12 58.45 32 304
13:19:26 stats
13:19:26 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 139.83 886.85 284 12 63.49 32 304
13:19:27 stats
13:19:27 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 140.79 864.42 284 12 74.21 32 304
13:19:28 stats
13:19:28 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 142.70 860.47 284 12 69.07 32 304
13:19:28 stats
13:19:28 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 141.69 904.89 284 12 55.28 32 304 Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
It shouldn't matter honestly, a 32 slot server should run at 66.7 no matter what map it's on. We have a TF2 32 slot open map server on an old Q9450(2.66Ghz yorkfield) with a higher splitpacket setting than you and we still maintain 66.7 fps consistently. Maybe john can help you, or you can try switching to managed. Although from my personal experience on a VDS switching from centos 5/6 64-bit to managed linux provided no difference. Unfortunately I think you lose your IPs switching from managed to un-managed.
If it's in your budget, try purchasing the 1270 machine here at NFO, for $200 it is a great deal. That's how we solved our problems.
EDIT: If there are any 1270v2 machines available, maybe you can try to switch to one of those on your vds. If turbo boost isn't kicking in on the machine that your vds is on, the 600Mhz boost would probably help a bit.
If it's in your budget, try purchasing the 1270 machine here at NFO, for $200 it is a great deal. That's how we solved our problems.
EDIT: If there are any 1270v2 machines available, maybe you can try to switch to one of those on your vds. If turbo boost isn't kicking in on the machine that your vds is on, the 600Mhz boost would probably help a bit.
Not a NFO employee
Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
Goldrush yields same results:
Code: Select all
13:55:53 stats
13:55:54 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 141.57 679.05 320 13 47.26 32 341
13:55:54 stats
13:55:54 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 141.14 696.85 320 13 45.43 32 341
13:55:54 stats
13:55:55 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 141.19 716.94 320 13 39.27 32 341
13:55:55 stats
13:55:55 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 143.23 743.14 320 13 35.25 32 341
13:55:56 stats
13:55:56 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 143.85 725.21 320 13 57.40 32 341
13:55:56 stats
13:55:57 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 144.58 686.29 320 13 56.89 32 341
13:55:57 stats
13:55:57 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 144.84 681.39 320 13 49.76 32 341
13:55:58 stats
13:55:58 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 145.45 682.10 320 13 47.14 32 341
13:55:58 stats
13:55:58 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 146.14 689.47 320 13 53.45 32 341
13:55:59 stats
13:55:59 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 145.48 693.93 320 13 46.42 32 341
13:56:00 stats
13:56:00 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 145.70 709.39 320 13 42.82 32 341
13:56:00 stats
13:56:00 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 144.65 723.73 320 13 56.02 32 341
13:56:01 stats
13:56:01 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 143.65 718.47 320 13 41.24 32 341
13:56:01 stats
13:56:01 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 142.89 733.92 321 13 63.19 32 341
13:56:02 stats
13:56:02 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 140.90 733.72 321 13 51.83 32 341
13:56:03 stats
13:56:03 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 139.09 711.99 321 13 42.46 32 341
13:56:03 stats
13:56:03 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 141.23 677.53 321 13 67.74 32 341
13:56:04 stats
13:56:04 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 140.64 677.08 321 13 53.70 32 341
13:56:04 stats
13:56:05 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 140.80 696.22 321 13 70.15 32 341
13:56:05 stats
13:56:05 CPU In (KB/s) Out (KB/s) Uptime Map changes FPS Players Connects
0.00 142.09 702.03 321 13 57.77 32 341
Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
soja wrote:It shouldn't matter honestly, a 32 slot server should run at 66.7 no matter what map it's on. We have a TF2 32 slot open map server on an old Q9450(2.66Ghz yorkfield) with a higher splitpacket setting than you and we still maintain 66.7 fps consistently. Maybe john can help you, or you can try switching to managed. Although from my personal experience on a VDS switching from centos 5/6 64-bit to managed linux provided no difference. Unfortunately I think you lose your IPs switching from managed to un-managed.
If it's in your budget, try purchasing the 1270 machine here at NFO, for $200 it is a great deal. That's how we solved our problems.
EDIT: If there are any 1270v2 machines available, maybe you can try to switch to one of those on your vds. If turbo boost isn't kicking in on the machine that your vds is on, the 600Mhz boost would probably help a bit.
Thanks for the response Soja,
I use to run the netsplitrate at max value but recently lowered it while testing/troubleshooting. I didn't notice a difference from 1048576 to 50000 so i just left it at that.
The IP address is key to me. I'm always scared of changing it especially after spending the past 1.5 yrs with it to build up the community and player base to have the server populated 24/7.
I was previously on the Intel X5690 which compairs pretty well to the 1270v2 and was experiencing the same results.
http://www.cpubenchmark.net/cpu.php?cpu ... 40+3.47GHz
http://www.cpubenchmark.net/cpu.php?cpu ... 40+3.50GHz
$200 is getting pretty steep IMO.
Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
If passmark takes advantage of multi threading(it looks like it does) then the 1270v2 beats it by a long shot(check the sandy bridge 6 cores on that page).
Bottiger overloads the crap out of his machines, out of pure speculation most of his locations are 1230v2's and he has 12-14 servers on each of them, at a mix between 32 and 24 slot.
I would never switch IPs either, but the only way to go to a managed vds would be to swap IPs(john correct me if i am wrong).
EDIT: the split packet setting doesn't take into effect unless you get to the threshold you set. Think of it as your rate settings, rate 50000 is around 50KB/s allowed, net_splitpacket_maxrate 50000 allows up to 50KB/s before you start to see choke, a stock TF2 server runs fine at 50k. A higher splitpacket means more network transfer in spikes, which means more data compression, which means higher CPU usage.
Bottiger overloads the crap out of his machines, out of pure speculation most of his locations are 1230v2's and he has 12-14 servers on each of them, at a mix between 32 and 24 slot.
I would never switch IPs either, but the only way to go to a managed vds would be to swap IPs(john correct me if i am wrong).
EDIT: the split packet setting doesn't take into effect unless you get to the threshold you set. Think of it as your rate settings, rate 50000 is around 50KB/s allowed, net_splitpacket_maxrate 50000 allows up to 50KB/s before you start to see choke, a stock TF2 server runs fine at 50k. A higher splitpacket means more network transfer in spikes, which means more data compression, which means higher CPU usage.
Not a NFO employee
Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
Python, I'm going to try switching you to the new Xen. You can expect to see that tomorrow morning, if all goes as planned with further testing today.
Re: VDS + TF2 + 32 Players -- sv_fps 66.7 Anyone?
When i initially setup the server with default settings i has having huge lag spikes, i would have blank sections in the net_graph the resolution which fixed that problem 100% was the net split packet cmd. I maxed that sucker out and have been running it maxed for the past 1.5yrs (with the exception of the past week while troubleshooting the fps issue).
Should i keep it at 50000 or do you recommend any other value.
While Bottiger might max out his boxes they do appear to hold a solid fps which is 100% related to cpu consumption.
The main reason why i'm starting to think all my issues are from VPS overhead is because i rent some dedicated servers which are some dated dual E5440 @ 2.83GHz & L5420 @ 2.50GHz processors. Each box has 8 core sand they handle a 24 man TF2 server no problem per core. That's 16 servers on two boxes no problem with a WAYYY shittier processor than what i have with NFO.
I'll wait for John to chime in on this conversation, see if he has any input/suggestions/recommendations on NFO's side of things.
Should i keep it at 50000 or do you recommend any other value.
While Bottiger might max out his boxes they do appear to hold a solid fps which is 100% related to cpu consumption.
The main reason why i'm starting to think all my issues are from VPS overhead is because i rent some dedicated servers which are some dated dual E5440 @ 2.83GHz & L5420 @ 2.50GHz processors. Each box has 8 core sand they handle a 24 man TF2 server no problem per core. That's 16 servers on two boxes no problem with a WAYYY shittier processor than what i have with NFO.
I'll wait for John to chime in on this conversation, see if he has any input/suggestions/recommendations on NFO's side of things.