@chon It's Jun 3, 2023. After three years, I'm going to finally do the migration to a new Dell server hardware. I'll update this thread for those who are interested. thanks.

@olivierlambert I watched this video by Tom @ Lawrence Systems. Now I understand everything.

https://youtu.be/1tJZAc-A4kU

I hope this video helps everyone else confused about backup.

@borzel Indeed, the command

"xe vm-cd-eject --multiple" worked! All VMs now correctly auto boot on reboot/startup!

@Danp they are stored in a FreeNAS physical server connected to another physical hardware with xcp-ng. The link via iScsi.

Thanks. I guessed that I needed to put the full path on FreeNAS as there is not masking when expose as NSF by FreeNAS. So I put the full path as "/mnt/MybooksVolume/VM_ISOs" and that seems to have fixed the problem.

@Andrew So I need to run xcp-ng 8.3 in order to get vTPM.

For those still running xcp ng v8.2.1 in production, would you recommend upgrading production to 8.3beta1? From the xcp-ng post, it seems to be very solid with some outstanding issues?

Also, xenserver is also available. Any thoughts on interoperability of Xenserver and xcp-ng and XO from sources?

I'm sure that by now, we should have Windows 11 vTPM support (& guest tools) on xcp-ng and Xen Ochestra, right? The last thread I found about this topic was from 2021.

I have experienced a catastrophic disaster while I tried to upgrade my rack servers. At first it worked very well by using the migration of VMs to the new server then assigning the new server as the master. Then, I decided that it was time to change the DHCP assigned address to take the place of the static ip , replacing the old server. Then all hell broke loose.

the xsconsole no longer saw the pms, yet the vms were still running!

decided to reboot the server to see if it's xsconsole not reading something and a reboot would solved it.

That's when I realized that all my metadata were gone! the vms were gone. the storage repositories were gone.

Yet, I know that the iSCSI repository, which stores all the vms, was still intact by looking at the mount of free vs. total space as presented in TrueNAS Core.

After many tries, I was able to reattach the iSCSI repository to the new server running, of course, xcp-ng. But there were no VMs presented in presented in a new instance of Xen Orchestra (CE). I thought that if there's a way to restore the meta-data for XO config and XO backup, I would be able to restore everything. The data has been downloaded off TrueNAS and into my MacBook Pro. It seems that these strategies aren't working.

I googled around to find out best practice or solutions to no avail.

I hope that you can help me please! These VMs are business critical. If someone can help me, the business won't be burning come Monday morning.

@chon It's Jun 3, 2023. After three years, I'm going to finally do the migration to a new Dell server hardware. I'll update this thread for those who are interested. thanks.

Hi there,

I am checking to see how loaded my servers are. When go to Dashboard >> Statistics and point to show me the "load", I'm shown mostly yellow with rare patches of green blocks.

When I look at the hosts themselves at Home >> Hosts >> Stats, the graphical CPU visualization, shows the CPUs are all running under 12% usage.

I've created a video to show you.

@olivierlambert Thanks. I understand Dom0 and VM0 principles. I didn't know that exited in
xsconsole, resulting terminal being a VM.

@Andrew If I am understanding your reply on dom0 correctly, does this mean that when I into the xcp-ng terminal and type htop and lscpu and they are both reporting 8 cores instead of 24, that's correct?

I have two identical servers. The other server, reports the right number of cores though.

How do I reconcile the difference between the two servers?

The only difference between them is the amount of ECC RAM (24GB vs 64GB) but I doubt that it would make a difference on this topi c.

Hi,

There is a weird issue happening because the number of cores reported between XO, Xcp-ng and lscpu + htop are different.

XO and xcp-ng report that there are 24 logical cores. Great. Also the BIOS sees 2 physical CPUs x 6 physical cores = 24 logical cores.

However, when I run lscpu and htop from the command line, they report that there are only 8 logical cores or 4 physical cores.

Has anyone bumped into this issue before?

are there 24 or 8 cores?

Does anyone have experience with nVidia Tesla P4 with XCP-ng with XO and Plex Media Server running in a VM with transcoding enabled.

@danp Yes, all my VMs are up-to-date.
Yes, I am running XO, the free and open sourced version.

In addition, this happens to all my 11 VMs.