Dedicated Server Performance Tuning

R

RamboTurnip

New member
I have a dedicated server setup on a Dell Poweredge running Server 2016, Xeon 2.2ghz and 32GB Ram, raid scsi, not the best but it is soley running 7dtd. Please can someone direct me to a document or guide for ensuring that 7dtd dedi server is running as best it can, I have noticed even when locally connected (client pc to server) the performance can be questionable, i.e. lagging items, kills, throwing a spear and pickup etc

There is nothing wrong when I do this locally on my main PC (100fps, no lag), or connect to a server hosted elsewhere. It isn't internal network bandwidth, and as stated the server isn't running anything else, purely 7dtd, so I'm wondering if there are some config options I'm unaware of when running a dedicated server. 

Your assistance is appreciated.

Kind regards,

James

 
I bought a cheap ($25) 240GB Crucial SSD in there, improved game performance when installed on it by like 300%

 
Hi Kraegor, I've got Raid Scsi running SAS drives, I've checked disk access under load and I don't think it's that causing an issue, albeit I will check and log resource monitor just to be sure!

 
Your RAID SAS drives aren't as fast as a SSD, and you need a minimum of an SSD for hosting.  Also note that your CPU's are below min spec.

Leave the OS on the SAS drives, and go gram a couple cheap 120GB SSD's for $40 and RAID them. That will help a lot. You might also want to check into what CPU upgrades are available for that, as minimum spec is 2.4GHz. Even at that, I wouldn't recommend going past 4-players with only 2.4GHz.

Lastly, you're going to see a noticeable improvement in performance if you run Linux over running Windows.

 
Ok taking on board the advice given , I've purchased an SSD and managed to source a 2.4ghz Xeon cpu which I'll install. With regards to Linux it's something I'll have to load into a vm and learn (Windows knowledge only here I'm afraid).

Is there any form of server performance check/benchmark within 7dtd dedi server that I can run for comparison before and after upgrades?

Thank you for your on-going assistance :)

 
If you use a VM you will be wasting some of the gained performance again and you need every bit. IF you use Linux, a dual-boot setup would be better. Any normal linux-distribution will leave the windows-install unharmed and just install a linux on some free SSD or HD space you point to and add a boot manager called "grub" that allows to boot into linux OR windows.

 
Last edited by a moderator:
Hi Meganoth, I'll only be using a VM to learn how to setup and use Linux, then install 7dtd dedi server, again to learn. Once comfortable with it, I'll blank the server and roll it out as the primary OS.

Again though, I'm assuming there are not inbuilt benchmarking tools or commands I can use to check how it's running?

Cheers

 
RamboTurnip said:
Hi Meganoth, I'll only be using a VM to learn how to setup and use Linux, then install 7dtd dedi server, again to learn. Once comfortable with it, I'll blank the server and roll it out as the primary OS.

Again though, I'm assuming there are not inbuilt benchmarking tools or commands I can use to check how it's running?

Cheers
Click to expand...
In the output_log of the server there are frequent lines where FPS are listed. From some comments by Fataal and my observations as rented dedicated server admin it seems this value should preferably be above 20FPS at all times. My old rented server (with an 8 year old Xeon Gold) was in the 20-30 FPS range normally and often dipping below 20 and it often made horde nights laggy.

My new rented server with current AMD server CPUs doesn't have a faster base clock than the old Xeon but its better IPC-value and better data path throughput seem to make a difference. My server FPS were suddenly in the 30ies and going down to 28FPS max. And we noticed definite improvements on horde night. And since we played a mod where a racing bike was double as fast as the normal motorbike the racing bike didn't result in massive tearing and delays anymore

 
I've just tested it at the moment, having changed to an SSD but waiting for the new CPU to come, I can see some improvement. I'm only using throwing the spear as a guage, and whilst it's still not as smooth as if I play locally, it has improved. I've set process priority to high, and albeit whilst I am seeing spikes up towards 30FPS, I suspect what you stated is a good judge for me to be able to tell if I'll see any in game improvement:

performance.JPG

 
I have 3 spinning drives in my server, one has the OS (windows 10), the second one has the 7 Days to Die dedi server installed, and the third drive has the save game on it.

I've had up to 6 players on it without any problems.

 
Last edited by a moderator:
Hi,
I just setup a little Archlinux dedicated server (Ryzen5 2400G, 16Go DDR4 RAM, SSD) and I ran into bad things in-game. Vehicle rollbacks, hard to kill zombies, like you.

Si I decided to customize my kernel by increasing the scheduler base frequency ( CONFIG_HZ ) from 300Hz to 1000Hz and everything is really better now.
I also got a better performance by setting up the cpu frequency governor from ondemand to performance.

 

 
So I finally got the updated CPU, and realistically it made little to no difference to the reported FPS:

image.png

It's a very odd state of affairs in so much as it's a lesser spec than what I used to run, indeed significantly worse as I ran multiple boxes on Hyper V in Alpha18 (same server without the SSD/CPU upgrades), and the only slow down you would witness would be during horde night. Under normal circumstances, throwing a spear would be fine.

All other operations seem fine up to now, the only thing I have left to do is reinstall with Linux and try again.

 
FPS in the server, is not the same as FPS in the client. A desirable server FPS is between 25 and 35. Any higher, and you're starting to waste CPU cycles, any lower and it's an indication of a bottleneck. You really cannot use it to measure performance otherwise as this has to do with clock cycles, and not frame rates.

 
I'm not sure what logs would be required to help diagnose it, I've made a comparison video below, the first section is localhost and second is on my server. I've only quoted the FPS as a judge based on feedback above, if there is more appropriate information to provide, or indeed use as a benchmark I'm happy to get it.

I do realise this in the general is considered a minor issue, however it makes using the spear near impossible.

Thanks

View attachment SpearThrow.mp4

 
You must log in or register to reply here.
Back
Top