Fps and Var on My Csgo Servers

[primofernando]

Hello,

I have a 3 core plan vds and i'm using debian 8 with the default kernel, and i have 4 csgo server, a competitive with 10 slots, a awp/aim server with 24 slots, a surf server with 24 slots and a minigames server with 24 slots, these servers work great. The problem is that when the 24 slots get around 20-24 players the sv fps and var become yellow and red when all players are alive and when i use htop on my ssh it shows that the task from csgo server that is around 20-24 players is using more than 100% cpu like 120% or more.

I'm using metamod and sourcemod latest version and i have some plugin like store, sourcebans and other small plugins.

I would like to know what can i do to optimise my vds because when the var is yellow or red the hitreg gets bad.

[stickz]

When using top do

Code: Select all

Shift + H

to figure out the CPU usage of the main thread. This will give a proper reading.

The latest 4.4 kernel in other Linux distributions (like Ubuntu 16.04) has sightly higher performance than the 3.x.x series in Debain 8.

Posting your ssh output of

Code: Select all

cat /proc/cpuinfo

would provide useful information as to the processor you're using.

Posting the output of

Code: Select all

sm plugins list

from server console may indicate if you're running any CPU-intensive plug-ins. Some intensive things to look for within source code include but are not limited to: OnGameFrame() and "Think Hooks."

Lastly, you have a little bit of an overlap with core to server ratios. I've always found servers + 1 cores to work best on VPS; except, in cases of games like L4D2 where 6 servers can very easily be stacked on 3 processor cores. Having more servers than cores generally puts you at a disadvantage as server CPU usage starts to climb; especially, with cases like 120%.

[primofernando]

When using top do

Code: Select all

Shift + H

to figure out the CPU usage of the main thread. This will give a proper reading.

The latest 4.4 kernel in other Linux distributions (like Ubuntu 16.04) has sightly higher performance than the 3.x.x series in Debain 8.

Posting your ssh output of

Code: Select all

cat /proc/cpuinfo

would provide useful information as to the processor you're using.

Posting the output of

Code: Select all

sm plugins list

from server console may indicate if you're running any CPU-intensive plug-ins. Some intensive things to look for within source code include but are not limited to: OnGameFrame() and "Think Hooks."

Lastly, you have a little bit of an overlap with core to server ratios. I've always found servers + 1 cores to work best on VPS; except, in cases of games like L4D2 where 6 servers can very easily be stacked on 3 processor cores. Having more servers than cores generally puts you at a disadvantage as server CPU usage starts to climb; especially, with cases like 120%.

Hello thank you for your response this is my htop with shift + H:


at the time in the first server(AWP/AIM Server) was playing 22-24 players

this is my cat /proc/cpuinfo output:

Code: Select all

processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 62
model name      : Intel(R) Xeon(R) CPU E5-2690 v2 @ 3.00GHz
stepping        : 4
microcode       : 0x415
cpu MHz         : 3000.096
cache size      : 25600 KB
physical id     : 0
siblings        : 3
core id         : 0
cpu cores       : 3
apicid          : 0
initial apicid  : 0
fpu             : yes
fpu_exception   : yes
cpuid level     : 13
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx rdtscp lm constant_tsc rep_good nopl eagerfpu pni pclmulqdq ssse3 cx16 pcid sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm xsaveopt fsgsbase smep erms
bogomips        : 6000.19
clflush size    : 64
cache_alignment : 64
address sizes   : 46 bits physical, 48 bits virtual
power management:

processor       : 1
vendor_id       : GenuineIntel
cpu family      : 6
model           : 62
model name      : Intel(R) Xeon(R) CPU E5-2690 v2 @ 3.00GHz
stepping        : 4
microcode       : 0x415
cpu MHz         : 3000.096
cache size      : 25600 KB
physical id     : 0
siblings        : 3
core id         : 1
cpu cores       : 3
apicid          : 2
initial apicid  : 2
fpu             : yes
fpu_exception   : yes
cpuid level     : 13
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx rdtscp lm constant_tsc rep_good nopl eagerfpu pni pclmulqdq ssse3 cx16 pcid sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm xsaveopt fsgsbase smep erms
bogomips        : 6000.19
clflush size    : 64
cache_alignment : 64
address sizes   : 46 bits physical, 48 bits virtual
power management:

processor       : 2
vendor_id       : GenuineIntel
cpu family      : 6
model           : 62
model name      : Intel(R) Xeon(R) CPU E5-2690 v2 @ 3.00GHz
stepping        : 4
microcode       : 0x415
cpu MHz         : 3000.096
cache size      : 25600 KB
physical id     : 0
siblings        : 3
core id         : 2
cpu cores       : 3
apicid          : 4
initial apicid  : 4
fpu             : yes
fpu_exception   : yes
cpuid level     : 13
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx rdtscp lm constant_tsc rep_good nopl eagerfpu pni pclmulqdq ssse3 cx16 pcid sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm xsaveopt fsgsbase smep erms
bogomips        : 6000.19
clflush size    : 64
cache_alignment : 64
address sizes   : 46 bits physical, 48 bits virtual
power management:

My Plugins List from my awp/aim server:

Code: Select all

  01 "SourceBans: Admin Config Loader" ((SB++) 1.5.4.6) by AlliedModders LLC, Sa
rabveer(VEER™)
  02 "Reserved Slots" (1.8.0.5908) by AlliedModders LLC
  03 "Basic Votes" (1.8.0.5908) by AlliedModders LLC
  04 "Server Advertisement" (2.5) by ESK0
  05 "[CSS/CS:GO] AbNeR ResetScore" (1.5fix) by AbNeR_CSS
  06 "SourceComms" ((SB++) 1.5.4.6) by Alex, Sarabveer(VEER™)
  07 "Chat Colors CSGO" (1.2) by Phoenix - Феникс
  08 "SourceBans++" (1.5.4.6) by SourceBans Development Team, Sarabveer(VEER™)
  09 "Map Nominations" (1.8.0.5908) by AlliedModders LLC
  10 "Server Redirect: Ask connect in console" (0.1.0) by H3bus
  11 "Player Commands" (1.8.0.5908) by AlliedModders LLC
  12 "RankMe" (3.0.0) by lok1
  13 "Fun Commands" (1.8.0.5908) by AlliedModders LLC
  14 "MapChooser" (1.8.0.5908) by AlliedModders LLC
  15 "SourceBans Checker" ((SB++) 1.5.4.6) by psychonic, Ca$h Munny, Sarabveer(V
EER™)
  16 "AFK Manager" (4.0.5) by Rothgar
  17 "Nextmap" (1.8.0.5908) by AlliedModders LLC
  18 "DR.API SHOW DAMAGE" (1.1.2) by Dr. Api
  19 "Basic Info Triggers" (1.8.0.5908) by AlliedModders LLC
  20 "SM SKINCHOOSER" (3.0) by Andi67
  21 "Basic Chat" (1.8.0.5908) by AlliedModders LLC
  22 "Basic Commands" (1.8.0.5908) by AlliedModders LLC
  23 "Admin Menu" (1.8.0.5908) by AlliedModders LLC
  24 "Rock The Vote" (1.8.0.5908) by AlliedModders LLC
  25 "Connect Announce" (1.6) by Arg!
  26 "Admin Help" (1.8.0.5908) by AlliedModders LLC
  27 "Fun Votes" (1.8.0.5908) by AlliedModders LLC
  28 "CS:GO Admin ESP" (2.2) by Root, Mitch
  29 "Easy Spawn Protection" (1.00) by Invex | Byte, based on work of cREANy0 and Fredd
  30 "Sound Commands" (1.8.0.5908) by AlliedModders LLC
  31 "Basic Comm Control" (1.8.0.5908) by AlliedModders LLC
  32 "Server Redirect" (0.2.6) by H3bus
  33 "Admin File Reader" (1.8.0.5908) by AlliedModders LLC
  34 "Client Preferences" (1.8.0.5908) by AlliedModders LLC
  35 "Server Redirect: Ask connect in chat" (0.1.0) by H3bus
  36 "MOTDgd Adverts" (2.3.5) by Blackglade and Ixel
  37 "Admin & PlayerTags" (5.0) by shanapu
  38 "Web Shortcuts CS:GO version" (2.4) by Franc1sco franug and James "sslice" Gray
  39 "GameVoting.smx"
  40 "SM Franug Player Colors" (1.3) by Franc1sco franug
  41 "Anti-Flood" (1.8.0.5908) by AlliedModders LLC
  42 "Simple Chat Processor (Redux)" (1.1.5) by Simple Plugins, Mini

And i have 4 servers to 3 cores but 1 server is only 10 slots and normally it takes 30% cpu not more so it's not a problem having one more server it almost takes no resources.

[stickz]

Hello thank you for your response this is my htop with shift + H:

Debain appears to work differently for enabling thread view. The image you posted doesn't have it enabled. It should look like this.
http://ask.xmodulo.com/view-threads-process-linux.html

this is my cat /proc/cpuinfo output:

NFO has faster VPS processors than an e5-2690v2 in some locations. You should ask if one (like an e5-2697v3) is available over a support ticket.

My Plugins List from my awp/aim server:

There's nothing which stands out in this plug-in list as being extremely CPU intensive.

And i have 4 servers to 3 cores but 1 server is only 10 slots and normally it takes 30% cpu not more.

Your screenshot shows approximately 178/300% CPU usage. Does it often get any higher than this?

[primofernando]

Hello thank you for your response this is my htop with shift + H:

Debain appears to work differently for enabling thread view. The image you posted doesn't have it enabled. It should look like this.
http://ask.xmodulo.com/view-threads-process-linux.html

this is my cat /proc/cpuinfo output:

NFO has faster VPS processors than an e5-2690v2 in some locations. You should ask if one (like an e5-2697v3) is available over a support ticket.

My Plugins List from my awp/aim server:

There's nothing which stands out in this plug-in list as being extremely CPU intensive.

And i have 4 servers to 3 cores but 1 server is only 10 slots and normally it takes 30% cpu not more.

Your screenshot shows approximately 178/300% CPU usage. Does it often get any higher than this?

Now i don't have my servers full, but yes there is an e5-2697v3 available, but it have less core speed 2.60ghz the actual processor have 3.00ghz. Does it have better performance even if it has less core speed?

And yes if all the servers that have 24 slots are full they can use 100% or more each.

[stickz]

Now i don't have my servers full, but yes there is an e5-2697v3 available, but it have less core speed 2.60ghz the actual processor have 3.00ghz. Does it have better performance even if it has less core speed?

The turbo is 3.6ghz on the e5-2697v3 and Haswell is faster than Ivy Bridge at lower clocks yes.

This link shows a 5% jump in performance. I've seen these results or better comparing those two NFO processors.
https://www.cpubenchmark.net/singleThread.html

e5-2970v2: 1,889
e5-2697v3: 1,981

And the cherry on top is faster disk rates and higher memory bandwidth.

And yes if all the servers that have 24 slots are full they can use 100% or more each.

Higher than 200/300% CPU usage could be your problem as well. You should upgrade or remove a server if you're experiencing CPU usage this high; as, you're running a definite overlap of more servers than cores.

The higher the overlap is the more usage from the top you have to allocate for your main server thread. It no longer goes 1 to 100% when an overlap is created because game-servers start saying "Give me 100% of this core or I lag".

This could be why your var msec is flooring down to yellow/red. If you're still experiencing this issue bellow 50% usage, in your case 150/300%; then, it's your game-server that is demanding too much on hardware.

[primofernando]

Now i don't have my servers full, but yes there is an e5-2697v3 available, but it have less core speed 2.60ghz the actual processor have 3.00ghz. Does it have better performance even if it has less core speed?

The turbo is 3.6ghz on the e5-2697v3 and Haswell is faster than Ivy Bridge at lower clocks yes.

This link shows a 5% jump in performance. I've seen these results or better comparing those two NFO processors.
https://www.cpubenchmark.net/singleThread.html

e5-2970v2: 1,889
e5-2697v3: 1,981

And the cherry on top is faster disk rates and higher memory bandwidth.

And yes if all the servers that have 24 slots are full they can use 100% or more each.

Higher than 200/300% CPU usage could be your problem as well. You should upgrade or remove a server if you're experiencing CPU usage this high; as, you're running a definite overlap of more servers than cores.

The higher the overlap is the more usage from the top you have to allocate for your main server thread. It no longer goes 1 to 100% when an overlap is created because game-servers start saying "Give me 100% of this core or I lag".

This could be why your var msec is flooring down to yellow/red. If you're still experiencing this issue bellow 50% usage, in your case 150/300%; then, it's your game-server that is demanding too much on hardware.

Ok thanks i will test the other processor and see how it behaves. I will post here if it works better.

[primofernando]

Now i don't have my servers full, but yes there is an e5-2697v3 available, but it have less core speed 2.60ghz the actual processor have 3.00ghz. Does it have better performance even if it has less core speed?

The turbo is 3.6ghz on the e5-2697v3 and Haswell is faster than Ivy Bridge at lower clocks yes.

This link shows a 5% jump in performance. I've seen these results or better comparing those two NFO processors.
https://www.cpubenchmark.net/singleThread.html

e5-2970v2: 1,889
e5-2697v3: 1,981

And the cherry on top is faster disk rates and higher memory bandwidth.

And yes if all the servers that have 24 slots are full they can use 100% or more each.

Higher than 200/300% CPU usage could be your problem as well. You should upgrade or remove a server if you're experiencing CPU usage this high; as, you're running a definite overlap of more servers than cores.

The higher the overlap is the more usage from the top you have to allocate for your main server thread. It no longer goes 1 to 100% when an overlap is created because game-servers start saying "Give me 100% of this core or I lag".

This could be why your var msec is flooring down to yellow/red. If you're still experiencing this issue bellow 50% usage, in your case 150/300%; then, it's your game-server that is demanding too much on hardware.

Ok thanks i will test the other processor and see how it behaves. I will post here if it works better.

It works i upgraded the cpu and now the var doesn't go to red or yellow and stays at low value thank you very much Stickz you're my hero.