Search results

for live migration with passed through devices, you must load a capable driver on the vf, in case of mellanox i think the driver was mlx5_vfio_pci then you need to adapt the config to have 'driver=keep' in the config in the 'hostpci' part like this: hostpciX: <pciiid-or-mapping>,driver=keep...

can you open the browser console before and then reproduce it? is there any error there?

I'm currently trying to get some of these and want to write a how-to guide, no promises if/when that arrives though. Also I'm not sure if it works at all with our current supported kernels, have to test first.

if you want 4 tapes in total, you'd have to lower the retention to 20 days i think, since each media set will only be 'finished' when the new one starts, so the first is now 21 days old

probably the node pdm connects to is one of the older ones, so that's probably why it won't work, but also don't detect it correctly. we know of this and i think someone is already working on improving this

hi, thanks for posting the output. The issue is that when encountering large values (such as the total of your nfs) perl sometimes converts integers to floating point values like this: And our rust parser expects an integer and fails. i sent a patch on the pve side to fix this...

hi, there is no compatibility matrix currently, but migrations from pve8 to pve9 should be working just make sure that both are fully up to date (with respect to the major version) can you post the output of pvesh get /nodes/NODENAME/storage --enabled 1 --format 1 --content images...

hi, is one of the nodes in the cluster on a slightly older pve version? can you post the version of that cluster? is the communication between the cluster nodes ok?

the guide is for pve but should be the same for pbs: https://pve.proxmox.com/wiki/Root_Password_Reset

hi, this is currently expected behavior, but you can simply remove the token from the pve ui. it's on our to-do list to remove the token as an opt-in mechanism, since the token may be used by other applications/clients too

Hi, not sure how you got into this situation, but the pve repository is not compatible with the remaining repositories: you have configured the 'bookworm' (debian 12) debian upstream repositories, but the 'bullseye' (debian 11) proxmox ve repositories first you have to determine what is...

i get what you mean, but it's hard to auto generate proper colors when the actual metric is a gradient, so we go from green to yellow to red, but since cpu and memory are on a scale, we have to fade between the colors to have some kind of difference for the most part. i'll see if we can set the...

yes, currently, you can open a feature request to make this changable for the @pam users too this is only true for @pam users. for those, the user has to already exist in the system for it to work, and the password is the system password for that user if you create a user in the @pdm realm, you...

do you have any firewall/proxy in between the vm and your workstation? (just to be sure: did you prepend the ip/domain with 'https' ?)

yep, this is a regression, i sent a patch today for this: https://lore.proxmox.com/pve-devel/04308263-4da1-4afc-8789-b868f5e49431@proxmox.com/T/#t

can reproduce the issue, but only when the datastore has an error. What does the content tab on the PBS gui show ?

it depends on what you want pdm to do. In essence, the pdm token for the pve cluster needs the same permissions for it's action as when you do it on pve directly. e.g if you want to start/stop vms, the token needs the VM.PowerMgmt permissions (or higher) for those vms. if you just want a...

thanks i could reproduce it here, and sent a patch: https://lore.proxmox.com/pve-devel/20251205091539.783566-1-d.csapak@proxmox.com/T/#u

did you clear the browser cache? was it the new datacenter bulk migration or the per-node one?

this was fixed very recently: https://lore.proxmox.com/pdm-devel/20251126131859.2745340-1-d.csapak@proxmox.com/