Higher server loads

[wrender]

After installing Cloud Linux fr om cPanels instructions here. http://docs.cpanel.net/twiki/bin/vie...ide/CloudLinux is there anything else that should be configured?

I tried editing the default CPU limit to 10, (It was 25). After doing that the server load went from about 1.0 to 4.2. Seems like something is wrong, I was expecting the server loads to drop if I drop the CPU lim it. Server load is running in a range of about 2.7 to 4.2 now. It used to run about 1.0 to 2.0

[iseletsk]

How many cores do you have? If you have less then 8 -- most likely it was caused by change in CPU limits.

Putting CPU limit very low can cause high CPU load. The reason being that load averages is proportionate to number of processes waiting for CPU, so if you start to slow down processes (which we do, to prevent them from going over limit) -- you will see higher load averages (even if you have plenty of idle CPU available)
We actually changed load averages, so it counts one LVE for at most one in load averages, so you need at least one process in 3-4 LVEs waiting for CPU to get to that load.

The general rule of thumb: On servers with 4+ cores -- use 1 core (ncpu=1, cpu=25) as a limit.
On servers with 1 or 2 cores, use 1/2 core per customer
or for 1 core servers, cpu=50%
for 2 core servers, cpu = 25%

[wrender]

Thanks for getting back to me right away. Im running a vmware server that is running on vmware cloud enterprise. I am renting 2 CPUs but I believe the vmware machine looks at it as if it were two physical CPUs. Would you be able to make a recommendation on this? Here is the results of cat /proc/cpuinfo

processor       : 0

vendor_id       : GenuineIntel

cpu family      : 6

model           : 37

model name      : Intel(R) Xeon(R) CPU           X5650  @ 2.67GHz

stepping        : 1

cpu MHz         : 2659.999

cache size      : 12288 KB

fpu             : yes

fpu_exception   : yes

cpuid level     : 11

wp              : yes

flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr

sse sse2 ss syscall nx rdtscp lm constant_tsc ida nonstop_tsc arat pni ssse3 cx16 sse4_1 sse4_2 popcnt lahf_lm

bogomips        : 5319.99

clflush size    : 64

cache_alignment : 64

address sizes   : 40 bits physical, 48 bits virtual

power management: [8]

processor       : 1

vendor_id       : GenuineIntel

cpu family      : 6

model           : 37

model name      : Intel(R) Xeon(R) CPU           X5650  @ 2.67GHz

stepping        : 1

cpu MHz         : 2659.999

cache size      : 12288 KB

fpu             : yes

fpu_exception   : yes

cpuid level     : 11

wp              : yes

flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr

sse sse2 ss syscall nx rdtscp lm constant_tsc ida nonstop_tsc arat pni ssse3 cx16 sse4_1 sse4_2 popcnt lahf_lm

bogomips        : 5319.99

clflush size    : 64

cache_alignment : 64

address sizes   : 40 bits physical, 48 bits virtual

power management: [8]

I have also modified my extracpus=-2 as described here.. https://helpdesk.cloudlinux.com/inde...up-are-running

[iseletsk]

You don need that, you should put extracpus=0 back.

I would recommend setting cpu to either 25 or to 50.
10 is way too low for two core server (and it is seen as two core/uses two cores from cln perspective).
Such low percentage will make all sites much slower, and will artificially inflate load averages.

[wrender]

Thanks Igor. I have set the extracpus back to 0, and set the default CPU back to 25%, as well as any users I have modified.

Server load is now running at 0.10!

Are there any other similar guidelines I should be concerned about for the memory limiting? For example I have set a limit to 512M, my server has a total of 8GB of memory.

My php.ini memory_limit variable is set to 256M

[iseletsk]

Usually -- the default limit swill work perfectly.
Depending on php settings & bunch of other things -- 512MB might not be enough (and if you see customers complaining about error 500 -- this is it).
I would recommend setting it to 1GB, might be even higher, given that you have 8GB of physical memory.

Basically -- what you want, are your limits high enough that most customers never hit them (if they do -- there is a chance of support request coming in), yet low enough that if one customer tries to abuse the system -- no one else notices.