@olivierlambert thank you. I am on the official channels since I pay for premium. I will wait to see the fix this week.

Also, same behavior as mine. I can see stats as well. I will report this on the Github site.

@olivierlambert I 100% agree its vague. I even told the person that. Like I was saying I was more looking for anyone who has ran a cluster in those other scheduler settings to get some feedback on it.

I guess being a little more specific, would socket or Core be better for VMs that are NUMA sensitive? Such as database servers or the like?

@olivierlambert

It "works" but we have more than one person that complain of "its not as snappy" as something like VMware or Proxmox (I know Proxmox is KVM).

I also have an opportunity that I am trying to get to use XCP but this is the argument I am getting. They say it performs like a vm on VMware with out the VMware drivers.

So I wanted to just get an understanding of what other people have seen when changing this.

@olivierlambert

So does the fact that tapdisk is single threaded beg to question that its performance is going to be tied to the speed of the CPU?

Also, when looking at some settings on some of the VMs created, I see on the VM this:

Where would I find what "default behavior" is?

I know this topic has been brought up and discussed a few times over the years, but I would like to get some insight into, in general, would be the best setting for performance and sclability for XCP in a "private cloud" hosted use case. In this use case, we have very little control over how our customers use XCP. There could be database servers, web servers, file servers, even Minecraft servers. We are currently just using defaults (CPU) and was wondering if moving to CORE would be better over all. We want to make sure we address things such as NUMA boundaries and things like that.

Any guidance would be appreciated or real world examples. Thank you as always.

@rfx77 @olivierlambert

I know this thread is over a year old, but I has there been any improvements to tapdisk over the course of this time?

@gduperrey

Thank you for the reply. The use case is to centralize all of our reporting and monitoring for all of our devices. We have a number of XCP host, various servers, etc, that we are responsible for, and all of them besides our XCP nodes are being monitored for real-time analytics by Netdata. We are also using Zabbix for trending and long term usage but would like to be able to move the XCP nodes into our Netdata cloud versus being just on XOA/XCP.

I also understand this would be overidden by updates and that you guys cant support it. I mainly wanted to see if it would "break" anything you already have in place. As a feature request, maybe have a way that you can report to a Netdata cloud instance. That would be nice.

Hello,

I know you have the Netdata binaries on XOA/XCP with the advanced analytics, but do you guys "support", and I use support as in not Vates supporting me on its use, but if I installed this, would this cause an issue with your implementation of Netdata? Also, would this need to be reinstalled after every update of XCP-NG.

Thank you for the help.

And if I am not mistaken, this setting in Proxmox has nothing to do with RDP connections, this would only apply to the "console" correct?

He is also comparing this to Proxmox which can be set to 256M for example.