@Bastien-Nollet I'm running c1e5f btw

@jerry1333 said: There is nothing else on that host and this is only host in pool but it's using 30% of cpu all the time? it's not using 30% of CPU, you see a graph of cumulated (switch is on) core consumption of your 32 cores. never switch this on. it adds up like that : 32x1%=32%, wrongfully letting you think you are at 30%ish CPU usage.

@Pilow perfect!

@Pilow said: @ph7 try to put RETENTION to 1 in the schedule, as you are using LTR parameters I was running the shedule with 20 mins backup retention and it is running fine together with the LTR And here is the manual I ran at 14:56 removed [image: 1775140196713-screenshot-2026-04-02-at-16-28-13-backup.png]

@rzr Great. Thank you!

@florent I can confirm that this fixes the issue!

@benapetr configure dhcp\static mode.

@Pilow The backups kept by LTR are just regular backups with a specific tag, which doesn't change how we treat them. If you want to avoid each of your LTR backup to depend on one another, we recommend to set a full backup interval value to your backup job, which will regularly force a full backup. (even without LTR, having an infinite chain of backups can cause problem in the long term, especially if no healthchecks are made)

@pierrebrunet Thanks for the update. Glad to know its not something unique to our environment and you were able to track down the cause!

@olivierlambert because xo-dsik-cli reviewed in last updates is not present and the github tells us to npm install here : https://github.com/vatesfr/xen-orchestra/tree/master/%40xen-orchestra/disk-cli

@TheNorthernLight Thanks for pointing that out, that totally makes sense, I should have realized to look it up that way. I guess I just think of XCP-ng as its own product so much that I forget it's based on Xen even though I know that.

@TechGrips Hi, Unused XVA means a XVA file is not linked to any metadata (to remove a backup, we first remove the metadata file linked to it, then we remove the XVA), it should be removed by clean vm step. It is not an issue, just an information. For the second warning "XVA might be broken", this is because the compressed file seems to be corrupted but we cannot be sure. We prefer to warn the user there may be an issue in the file. This should encourage to remake a backup.

That's indeed more on @Team-Hypervisor-Kernel side. Maybe a driver/firmware version mismatch?

Hello @tsukraw , normally your client should be able to bind the new license on their XOA. BTW, you talk about binding the license to a pool but an XOA license is bound to an XOA. I suggest your customer open a support ticket so we can help them with that issue.

@DustinB thank you! I suspected as much, but as my wife points out all the time, I am terrible at finding things, so I figured I would ask.

Support tunnel is active but I have troubles to access support "Oops, something went wrong! Don't worry, our team is already on it". I'll keep trying, thanks.

Adding @Team-XO-Backend in the loop

Nice catch @vagrantin ! Feel free to open a new thread if you have other problems.

Ping @gregoire

@AshleyDe It was exactly that, but complicated by the abandoned status of the fuse-native module being used for XOA in 2023. That module contained binary blobs, but unfortunately no binary for aarch64. Being abandoned meant getting aarch64 into that module was not possible. My horrible bodge was to swap to an alternative module for fuse-native that built from source. Thus it avoided the missing aarch64 blob, but required additional dependancies for the compile to work. I'm no longer using XOA on the aarch64 platform in my home lab so I haven't kept up to date on whether this is still needed in 2026.