Lot of assumptions in that The Register write up.

Hobbyists not wanting to pay for licences, running on old hardware, supporting charities, etc. (Paraphrased)

The reality is that any "hobbyist" playing with this kind of tech, it's really a hobbyist. I'm not sure I would class anyone playing with this kind of technology a hobbyist, this suggests a lack of professionalism towards these technologies.

Let's be honest, 95% of people playing with this tech are going to be in some fairly high end jobs and/or high end IT consultancy. No-one has a hobby to use Type 1 Hypervisors, server grade hardware (brand new OR second hand) and commercial grade networking infrastructure required as a baseline to get this set up.

Many people in the IT world have probably never even heard of the concept of a Type 1 Hypervisor, let alone want to get into this well-known "Hobby".

The majority of the community contribute to progress their own understanding from a professional perspective and to pay it forward to support those who are chomping at their heels.

The reality is that for most big corporate enterprise companies (Citrix and many others) is that they often go down the route of providing £x,xxx training courses to learn their tech and expecting organisations to pay for this. Many organisations don't invest heavily in their staff through corporate training programmes. So people are left to find these things out on their own time in many cases.

No individual outside of the corporate working world is going to pay this level of investment for a course that, in my experience (not related to Citrix specifically), often provides little added value beyond a few Google searches, a ton of questions in the community and a watching a few YouTube videos.

There is a big difference in a community of professionals and a community of hobbyists.

If anything, that write up just shows how little they still understand about the community.

Good old Windows automatic upgrade/reboot/annoying-as-hell-feature has kicked-in.

I've just re-tested this since the above has happened (causing the endless disruptions across everything running on Windows). But anyhow.

Seems to be magically fine now from an XO perspective (I've not done a single thing since this post VS this annoying auto Windows reboot/upgrade);

No idea what this issue could have been caused by, highly likely a weird compatibility issue with a specific Google Chrome version is what I would come to the conclusion of.

Seems like the age old solution in IT has worked again..... Turn it off and back on again

I've had a few issues over recent years (other technology stacks, nothing related to XCP-ng stuff) where Google Chrome Version X,xx has broken something that should "just work", so I'd add this to the list of items that fall into that category based on this not working then magically working again after a forced reboot + general upgrade etc.

It certainly does explain why I was confused along the way Anyhow looks like this is solved.

What a weird issue, nothing related to XO in the end after all. The joys of working in IT/Technology/Software, so many layers inside and outside of control etc.

That would explain things, and also why my latest test using AlmaLinux 10 also failed in the exact same way that Rocky Linux 10 did.

I just wish I read this reply 12 hrs ago

Think I'll give this stuff a go on v9 of Rocky Linux and AlmaLinux to check things are working, which I imagine will work fine.

I've just confirmed the EPEL version running on AlmaLinux 10 at the time of writing this comment is indeed v 10.1

via the command;

sudo yum repoinfo epel

@dknight-bg What is the best way to keep track of this stuff proactively, i.e. What would I need to look at so that I could have determined what you have commented myself?

Commenting for my future self.

This works for CentOS Stream 10;

# Install Extra Packages for Linux, where xe-guest-utilities-latest lives
sudo yum install epel-release

# Install
sudo yum install xe-guest-utilities-latest

# Start the Service on Boot
sudo systemctl enable xe-linux-distribution

# Start the Service Now
sudo systemctl start xe-linux-distribution

# Check the Service is Running
sudo systemctl status xe-linux-distribution

# Check it's installed
sudo yum list installed xe-guest-utilities-latest

I wasn't particularly committed to Rocky Linux 10, so thought I'd give the usual steps a run on a different similar OS using CentOS Stream 10. Works absolutely fine.

As a comparison, when I run the above steps on Rocky Linux 10, it fails on the second step above when trying to install from yum, which appears to be due to this not being included in EPEL.

sudo yum search xe-guest-utilities

No matches found.

Curious to understand if anyone knows why EPEL is different between CentOS Stream 10 VS Rocky Linux 10. Is this just some upstream thing that isn't perfectly aligned?

Struggling to find a full and complete example of how to do this, and everything I've tried that has worked before and via docs I've found just isn't working.

I'm probably missing something fairly obvious for those who have done it before.

Tried via XCP-ng Centre (8.2, with all patches applied) - Seems to be some weird bug in there from what I've seen on the forums with the guest-tools.iso mount not working.

Tried via XOA with guest-tools.iso mount, and then via console to run commands - Which didn't work

Tried via XO from Sources with guest-tools.iso mount, and then via console to run commands - Which didn't work.

Followed guidelines from here https://docs.xcp-ng.org/vms/#linux-guest-tools

But it's not complete documentation, I think the missing bit of info is "enable the appropriate repository" which if documented, then it should mean that things can be installed via Yum with

yum install xe-guest-utilities-latest

But unsure the steps to take to configure something this ambiguous.

I'm also a bit confused why the guest-tools.iso isn't working like before. I can only assume that it's because Rocky Linux 10 is newer than XCP-ng Server 8.2, so it's probably an alignment issue.

What am I missing to get this working?

Also curious what the use case of this is.

The only genuine use case I can think of for this is for a basic Storage Repository containing ISOs for VMs so they can spin up.

Or are you expecting an AWS S3 equivalent or Windows File Explorer equivalent?

Thanks @lawrencesystems - I've finally got around to watching that video, explains the content well, quite interesting to see that happening in real time when the RAM is being land-grabbed by other VMs when things are getting full based on the different configuration settings.

@tjkreidl That sounds like it's answered 99% of initial questions to be fair.

The 1% I'm not clear on is, what would happen in the scenario whereby say I have 10x VMs, which in total took up 100% of RAM for example - but they were all turned off.

Would I still be able to create an 11th VM with 100%+X RAM allocation and still turn it on successfully? i.e. because the other 10VMs are turned off?

From what you've said, I'd come to the conclusion of 'yes'.

Seems easier to have a conversation around the topic rather than me trial and error playing around blindly.

How does this work? I've yet to reach the max CPU + RAM allocation to the VMs on the XCP-ng Server Host, so I'm just thinking ahead.

For each VM, are the CPU and RAM allocations fixed, i.e.

1. 1x CPU Core and 2GB RAM, is that 100% dedicated to that VM, even though it may only be using 10% of that capacity and/or may even be turned off.

or

2. Is the capacity shared in one way or another

or

3. A mixture of both depending on some configuration settings I haven't spotted yet?

I'm pretty sure it's #1 from the above out of the box, but just checking.

And if my current understanding is correct, what options are there for over provisioning with the acceptable consideration that performance degradation is likely.

In reality this is for a lab setup, so this is more to over provision to allow more playground VMs as more often than not only a small handful will be turned on at any one time.

Which thinking about this just as I type raises another thought. Is it possible to over provision as long as things are turned off? That's a bit of a second question I guess as both scenarios are valid.

Good old Windows automatic upgrade/reboot/annoying-as-hell-feature has kicked-in.

I've just re-tested this since the above has happened (causing the endless disruptions across everything running on Windows). But anyhow.

Seems to be magically fine now from an XO perspective (I've not done a single thing since this post VS this annoying auto Windows reboot/upgrade);

No idea what this issue could have been caused by, highly likely a weird compatibility issue with a specific Google Chrome version is what I would come to the conclusion of.

Seems like the age old solution in IT has worked again..... Turn it off and back on again

I've had a few issues over recent years (other technology stacks, nothing related to XCP-ng stuff) where Google Chrome Version X,xx has broken something that should "just work", so I'd add this to the list of items that fall into that category based on this not working then magically working again after a forced reboot + general upgrade etc.

It certainly does explain why I was confused along the way Anyhow looks like this is solved.

What a weird issue, nothing related to XO in the end after all. The joys of working in IT/Technology/Software, so many layers inside and outside of control etc.

Interesting.

Chrome with a Dark Mode theme installed;

Edge (default as I don't use);

I can only assume that it's the Dark Mode theme I've installed on Google Chrome getting in the way with things and getting a bit confused.

No wonder why I was getting confused, I couldn't see the thing I was expecting to see, yet it was right in front of my face hiding in plain sight

@Andrew "You can access the VM console from XO as long as XO has server (not guest) access. It's out of band access to the virtual VGA console."

How would I verify what level of access it has? Is there a page/screen in XO that shows this info?