Question PHP CLI thread on Multi-Core VPS

[WebHostingAce]

Hi,

I have 2 VPSs from a same hosting company.

VPS 1,
16 Core
64GB

VPS 2,
4 Core
8GB

I can run a PHP script via CLI about in 10mins on VPS 1 but VPS 2 takes about 30mins.

How could be possible?

As far as I know PHP CLI Script runs on a Thread/vCore. So the both VPS 1 and VPS 2 should have completed the task on same time limit.

Both VPS are within the Memory Limit when running this script. Also the Disk Read/Write Speeds.

Anyone got any idea?

Thank you.

 

[Ohya]

Hi

Is complicated to say, buy maybe you have any limit in your VPS from node, ask your vps provider

But VPS 1 maybe faster in any way if you have some querys to database

But in any case, this isnt a plesk issue in any way, imposible to resolve it from here.

 

[WebHostingAce]

Hi Ohya,

Thank you very much for your reply!

I think we can discuss about server related topics in 'Open Topics' section.

It the same website on same VPS cloud.

 

[Ales]

It's quite possible to end up on very different hardware nodes even if both VPS'es are from the same provider and even if they both use the same virtualization type. Sure, they are saying it's a cloud, but in reality, it's just a cluster of interconnected servers so the actual hardware might be different, underlying limits might be set differently, etc.. The customer count on the hardware nodes might be noticeably different too.

I'd run some generic benchmarks on both to see what's going on...

BTW, some types of virtualization will allow you to see the actual CPU type, others will mask that information... Still, doesn't hurt to try a simple 'cat /proc/cpuinfo'...

 

[WebHostingAce]

Thank you Ales!

Provider is - Vultr

I'm not sure how they distribute computing power according to number cores.

PHP CLI Script seems to run on one core with 1 PID.

 

[Ohya]

I know vultr just run one vps there, is like digitalocean not is real cloud I think is just a baremetal servers nodes, but not shure

In any case check if you have same CPU model in both, but in any case as always virtualized system can have overload nodes and may be no cpu guarranted.

Is difuicult to say, with low prices, as your provider.

 

[Hex]

Still, doesn't hurt to try a simple 'cat /proc/cpuinfo'

Hi @AusWeb
Since I'm dealing with Vultr
Can you post the output from both servers? I'm curious

cat /proc/cpuinfo

 

[WebHostingAce]

Hi @Hextrator

Please find the attached Zip file. Within that Zip file you will find 2 txt files.

 

[Ohya]

May you can test your performance with some stress test, stop all your web services and database for mantaince and try stress your CPU, core usage, 2 cores usage etc and compare performance,

Also is cheap to deploy new instance in vultr for some hours and test if you get same issues in clean server with out load ( this test may help if you cant go down for mantaince ) just for test infraestructure and see if you can reproduce this in similar servers