Hi all,

 

I have since 2023 a Tower NAS running unRAID 6.12.10. The specs of this NAS are as following:

 

CPU w/iGPU = AMD Ryzen 7 5700G
Mobo = Gigabyte B550 AORUS ELITE V2
Dedicated GPU = MSI GeForce GTX 1050 Ti 4G OC
Case = Cooler Master Force 500
RAM = 1x G.Skill Aegis F4-2666C19D-16GIS (DDR4 16GB) & 1x G.Skill Value F4-2666C19D-64GNT (DDR4 64GB)
PSU = Corsair CX550
Cache (zfs pool) = 1x Samsung 870 QVO 1TB & Samsung 870 QVO 4TB
Array (xfs)= 3x WDC 3TB & 1x WDC 4TB & 3x WDC 8TB & 4x Toshiba 18TB

But now I am running 21 docker containers for 24 hours a day, 7 days a week and 365 days a year.
And since I'm using that amount of docker containers, a lot of the webUI's of the docker containers are reacting slow.
First the webUI of unRAID itself was also slow and not reacting, so I've restarted Nginx, and that solved the slow webUI of unRAID, but still most of the webUI's of the dockers are so slow...
I must mention that while using the dockers webUI's, I use Google Chrome (Version 129.0.6668.60 (Official Build) (64-bit)) and PIA as VPN, with "Allow LAN traffic" enabled.
 

I also heard about the CPU pinning for docker.
But the ones that told me about it, said they don't have the knowledge how to set it, to get the maximum CPU power for the containers.
What I did was I left 4 threads for the system and the other 12 threads assigned to all the docker containers.
Below you'll find the screenshots, to clarify things.

The loads of the server, measured with htop, at this time are all above 30, and most of the time stay in the 30's or even higher.
And at the moment I'm not actively using the server or the containers.
All that now is running on my PC is the unRAID webUI, some docker webUI's that I have open as always to check some of the dockers progress'.
I don't know if those load averages are very high or normal, but they look to me as if it is significantly higher than a normal load.

 

So now the questions:

 

What can I do to make the webUI of the docker containers less laggy, or to response in a quicker way?
I think that a fairly "good" and fairly recent CPU is installed.
Can I resolve above problem with an upgrade of the CPU that has about the same or less TDP?
As far as I know, this is one of the better AM4 socket CPU's, so upgrading would at least cost me a new motherboard and CPU...
Or is upgrading my CPU overkill, and a unnecessary spending of money?

 

I know that my physical RAM configuration isn't ideal, but does it also slow the system down, in comparison with only the 64 GB RAM installed?
I am saving some money right now, for a 2nd set of RAM (G.Skill Value F4-2666C19D-64GNT (DDR4 64GB)), but I cannot afford that right now.
Also I don't think I need more than the 80 GB at the moment, but currently the system is using 87% of the available RAM, mostly due to the ZFS pool, which is sucking up RAM like a old vacuum cleaner.
I deliberately created a cache pool with ZFS (ZFS Raid0 2 devices to be precise).
This is because I was crashing every week with my single cache disk, then with BTRFS and XFS...
Unfortunately, I was still using a PCI-e card with Marvell controller to increase the number of sata ports (to be precise, the Supermicro AOC-SAS2LP-MV8).
A month or two ago, I was advised on here to replace that card, which I of course did immediately, and then at least the cache problems were solved.
That was after I already had a cache ZFS pool in use.

Here the screenshots of my CPU pinning:

 

 

And here the screenshots of my docker containers:

 

 

Thanks for reading my never-ending story and sorry for this terribly long post, but I think I've told everything clearly that way.....
I hope you'll have some tips and advice for me!

What you've done with the cpu pinning is good, I've done this on my compute intensive servers so they are still usuable.

 

Docker page has always been the slowest, especially in advanced mode.
One setting in Docker settings was added in recent versions that disables a visual display script, this can make a noticeable difference to performance on Docker page.

Thanks for helping me out with the CPU pinning.
So that setting seems right.

I didn't mean my docker page was slow, but the webUI pages of my containers are slow...
Also that's why I added the screenshots, so that you'll can see which containers are running now.
Maybe it was all a bit confusing as to what exactly it was about.

My docker page was indeed slow, but so were the other unRAID pages.
That was easily fixed, by restarting Nginx via Putty.

 

At this time, I'm not running the poste.io container anymore.
So that gives 20 containers that are running 24/7/365.

Edited September 30, 20241 yr by rikdegraaff

I'm sorry for posting twice, but I don't understand why nobody has responded to my topic, except for @tjb_altf4.
I don't know if double posting is allowed on this forum, but from my experience on other forums, it's usually not appreciated or even allowed.

 

I asked about this before on Reddit, and they said to try it here.
So, I did, but I still don't have all the answers.

 

Is anyone else able to help me with these questions?