@Rod-G These have 9.4.2-xxxx and I need to go through and update everything now that I see how far behind I am. the second VM completed fine after its reboot, something was just stuck in a dirty state and got in the way of the import or health check.

@dthenot Thanks a lot for the information. I did some more testing on my end and I've now noticed differences in handling VDI format across different SR types (local LVM and EXT). Should I expect even more differences across remote SR types like LVMoHBA or NFS or are these differences more like block based vs file system based SRs? Any way, it looks like I have to consult the following keys: image-format vdi_type type In my tests, local EXT SRs tend to have only one of them, while local LVM SRs tend to have first two or all three, depends which one was used when creating the VDI.

Thank you all (@olivierlambert , @teddyastie ) for your valuable responses

@tonyp90 great, thanks !

Closing the loop on this one — VSA-2026-021 went up yesterday (June 10) covering CIFSwitch / CVE-2026-46243: https://docs.vates.tech/security/advisories/2026/vates-sa-2026-021 A few things worth flagging for anyone following along: Severity landed at Moderate 🟠 — same ballpark as CopyFail/DirtyFrag, as Lucien anticipated. XCP-ng 8.3 and XOA both confirmed affected. XCP-ng 8.3 fix isn't in the main repo yet. The advisory notes there's a publicly available package with the fix, but it's not in the standard channel — Vates is asking people to reach out for the install procedure so you don't break future Rolling Pool Updates. So don't go hand-rolling the kernel commit yourself if you want to stay on the RPU path. XOA is already handled — fixed in Debian kernel 6.1.174-1, pushed via the unattended update mechanism. Just note the XOA VM needs a restart for it to take effect, and anything older than Debian 11/12 won't get the update and needs an OS upgrade first. Mitigation is unchanged from what we discussed: blacklist the cifs module if you're not using SMB-based SRs (which breaks SMB SRs, so only if you don't rely on them). Good turnaround given the disclosure-to-advisory window. Thanks again @LucienLassalle and the security team.

@Andrew I did suspect that would be sufficient, but we need to think at feature level, and as mentionned there is no such thing we could do "quickly" for linux and other OSes. I anyway did a brain dump of my investigation before posting my previous message and we do now have an entry in the roadmap for it, which was not the case previously.

@psafont Thanks for the quick response and clarification. I appreciate you opening a work item for this. Looking forward to seeing this improvement in a future release.

The same as @john.c and also XO Lite tends to be less a priority because less critical than the full fledged XO (the priority is to replace entirely XO 5 in the next releases). Why you would need XO Lite outside basic actions? It's mostly meant to bootstrap XO itself and do basic operations (which is already the case, at least with many basic features already). Initially, the goal hasn't moved: replacing XenCenter. We are moving in that direction, but again, I think it's more important to get XO 6 finished first. I'm curious to understand more the use case of XO Lite in your context @unreal-shizzle ?

Good to hear, thanks a lot for your feedback.

That leftover NIC is probably a stale PIF that XAPI is still holding, even though the card is physically gone, and the Refresh button re-scans rather than removing it. The docs have proper "remove a physical NIC" steps that end in forgetting the old PIF with xe pif-forget: https://docs.xcp-ng.org/networking/#remove-a-physical-nic. For the GPU side, the PCI passthrough flow (hiding the device from dom0, then assigning it) is here: https://docs.xcp-ng.org/compute#detaching-a-pci-device. If the GPU still won't show up as assignable once the stale PIF is gone, it might be worth a mention to @Team-Hypervisor-Kernel.

@danp I see your response to my concern was to add a block on the script being able to run on 8.2, without an update to the elif bug @anthoineb acknowledged. While I can remove your addition and fix that other bug myself, I just wanted to give you some feedback as to why I expressed the issue in the way I did - it wasn't because I feel an entitlement to anything, just sharing my perceptions to @antoineb I have been a xen cli user for a couple of decades, give or take. I moved to xcp-ng with a lot of PV legacy. That was a blocker to using 8.3 straight away, otherwise I would have. The workflow I had established to update the VMs to HVM included adding a new volume for /boot and /boot/efi, and then switching boot OFF on the legacy disk. It was worrying when I went to complete the migration to HVM with the announced EOL of 8.2.1, that suddenly the disks I needed to work with, both legacy and new, had disappeared in XO so I couldn't follow my SOP to move the VMs to HVM, and then proceed to upgrade to 8.3. Given how relatively widespread this issue seems to be, I am surprised that Vates wouldn't treat it as a reputational issue and want to help users to resolve it so that corporates can have confidence to move their cloud to xcp-ng platform, even if they are a startup or in the initial stages of a move and can't yet invest in Pro Support. I note you have a "Pro Support Team" badge, and understand that may not mean you work for Vates, but I gather offer Pro Support. Valuable service. While I am not encouraging this thread to become a marketing channel, it would speak well for xcp-ng, Vates and the Pro Support community if this important issue was handled differently - even if it was "we have tested on 8.2.1 internally and should be able to fix it for you quickly and cost effectively if you join our X plan or request a quote on Y page." I appreciate all you've shared and wasn't being dismissive of your recommendation to upgrade to 8.3, just the timing of this issue - arising right before the EOL of 8.2.1 made that a bit of a "chicken and egg problem". I gather from what you've said on this thread that you don't expect upgrading to 8.3 to be a problem with VDI in this state, so I guess I will just soldier on with creating an xcp-ng CLI SOP to upgrade the remaining PVs to HVM, upgrade to 8.3, and then use the script that has been shared - since you've now made it clear in the code that it shouldn't be run on 8.2. I am thankful that it was shared as guidance, if not a fix in my situation.

Maybe you can check for a bigger timeout on your UDM?

Hello. Thank you for your input. I am aware, that this is more of a warning message, than a error. I´m just trying to figure out, what is my best way to go here. My plan was: Setup a repljob for the vm in a lets say hourly interval On the day of the migration, shutdown the vm and start the last replication manually Disable the cr job Start the replicated vm on the new pool, check it and if all is ok, use it as new vm, otherwise start the old vm. The documentation says "If you want to start a VM on your destination host without breaking the CR jobs". Tbh i dont care about breaking the job. If everything works fine, i dont need it anymore, if not, i can setup a new job pretty fast. I was just wondering, if the new vm will stay in "blocked mode" for ever. Kind regards

@dinhngtu said: I've taken a quick look, looks like it'll be solved as part of the Windows guest agent overhaul, so please look forward to that. Thanks for the info. I will be looking forward to that, indeed.

@florent Thanks for your reply! We have started to migrate thousands of VMs, so disk transfer speed is important for us.. We will also do our detailed tests soon with different threads setting and publish it here. I think threads=1 is a good and logical default, but not efficient. Others might complain if you set it to a higher value. Configuration option would be a real good solution.

@pierrebrunet I inadvertently found more information yesterday. I installed an old e492b commit on a new machine, configured identically to the original, and got a different error but the same effect. The new error is: xcp xo backup Error: read ETIMEDOUT That lead me to this page: https://xcp-ng.org/forum/topic/10799/backup-timeout-error I moved the XO machine onto the same subnet as XCP, leaving the storage behind in a different network. This fixed the issue, at least on commit e492b ... I have not yet tried it on the new release... however I am pretty confident it will be successful. If I'm honest, I think this is a tuning issue between XO and XCP. XO is now mounting an NFS share across this subnet boundary and that mount is not having any sort of performance issue, but for some reason XO times out transferring the same data from dom0? That seems off to me. It's not a probem moving XO to the other subnet, so I've done that, but maybe this deserves a look in the future.

Because it works already better than the GO tool from Citrix… There's no urgent fix to do, I personally use it in my production since it's available. It's just less a priority for extra features because it's already ultra stable. Right now, we choose to work in priority on XCP-ng 9.0 than the Rust tools, we can't do everything at once yet.

I'm not positive, but this may be available RSN in xo-lite.

@tuxen Doing some research, it doesn't look like the Xeon's I have are affected. But I'm willing to try the next time I need to reboot. Will report back after that.

The rule is oddly written, and may conflict with another similar one that already exist in the distro (hence may not be useful to begin with). The modern generic rule for doing vCPU hotplug is, which would be preferable to the current z10-xen-vcpu-hotplug.rules. ACTION=="add", SUBSYSTEM=="cpu", ATTR{online}=="0", ATTR{online}="1"