So that explains it then Free BSD boots and since the kernel loads successfully the PV drivers, it's fine by us.

Excellent news! I suppose we should document it somewhere, ping @thomas-dkmt

Maybe because SM version used for XOSTOR is a bit different? @dthenot do you confirm?

It's a FreeBSD, so the PV drivers are bundled in the kernel. Maybe that's enough for XO to detects when they are loaded by populating the Xen store.

Let me ask @Bastien-Nollet to review a bit of the logic behind the tools detection

Hi,

I think this issue is being fixed soon. Adding @dthenot in the loop to be sure

Your error is simply a lack of updates on the host you want to add to the pool, nothing else

Hi,

Can you copy the result of xe vm-param-list uuid=<UIID of this VM> and paste it here?

So you have to dig in the SMlog to check what's going on

That would be interesting if a we have a test driver solving this very issue here, but I wouldn't expect too much either Those drivers are as bad as the chips

Yeah I would avoid those shitty cards as possible. Is there any more recent driver anyway?

As I said, you were lucky: it's a question of "not crashing" consumer grade hardware between a driver version and a firmware version (even on server-grade hardware it could happen, but it's simply less likely.

That's why I would advise to update first all the firmware first. The next step is to play with a alternative kernel driver for the NIC (if there's one), you could even start there if you like.

At least, you know how to trigger it (maybe a simple iperf would be enough) so you can test various drivers and/or firmware versions.

It's unrelated, are you really building a production infrastructure with non-server grade hardware? There's a good reason people don't use consumer-grade hardware: it's not meant for it, it works for basic usage but you can easily encounter buggy BIOS, firmware, ACPI tables and so on. It doesn't have the QA process done on server-grade hardware.

It's a LOT better to purchase refurb stuff (even a cheap refurb 10G Intel NIC will be 1000 times better than any RealTek crap you can purchase brand new).

Another system which is also a consumer grade motherboard, right?

Hardware name: ASUS System Product Name/PRIME Z790-P, BIOS 1663 08/08/2024

You BIOS is also outdated.

What NIC are you using in there?

If it worked with 8.2, it's potentially the version of the driver for the NIC, in general things can go well with specific firmware+driver version. Maybe you entered a different combo that's not great. So first, updating BIOS/firmware of the machine AND the NICs is likely the best next move (or swapping for a better NIC).

For the tools, @dinhngtu can provide guidance

Yes, the whole folder for each VM and job ID. I'm not 100% sure about the folder "on top" of it, but better safe than sorry

You can move all the files related to this job to another place (with rsync for example, or scp or cp)

Question for @TeddyAstie maybe

You have a rather long chain now, but it should coalesce anyway

Do you mean at VM creation or later in the VM lifecycle? Adding @lsouai-vates in the loop

Hi,

No, there's no reason XCP-ng patches are doing issues like this (at least I'm not aware with our customers and our own install is also patched too: no issues).

This makes me believe it's a configuration/environment issue on your end. I know it doesn't help, but at least it's XCP-ng unrelated