Recent content by jsterr

Checkboxes seems to be fixed with new proxmox-widget-toolkit:amd64 (5.1.6, 5.1.7) Start-Date: 2026-02-26 12:03:27 Commandline: apt-get dist-upgrade Upgrade: proxmox-widget-toolkit:amd64 (5.1.6, 5.1.7), corosync:amd64 (3.1.9-pve2, 3.1.10-pve1), libvotequorum8:amd64 (3.1.9-pve2, 3.1.10-pve1)...

Does VM console work when you check 4-6 Machines ?

Hello, on some training-vms I upgraded to PVE 9.1.6 and Im having some weird issues atm I cant track down. Im not sure if it is related to the 9.1.6 upgrade. * Ceph -> OSD -> Global-Flags: Ceph Global Flags Show Integers "0" or "1" instead of Checkbox ** this makes it hard to use and multiple...

Die Erstbefüllung würde mit 4 MB / s einfach zu lange dauern, daran hatte ich aber auch schon gedacht. Bezüglich Punkt 3) Offline Backup heißt für mich, kein Server mit den PBS-Pulls drauf sondern Backup ist komplett auf eigenem Medium und kann z.B. in Tresor gelegt werden (Beispielshaft)...

Hintergrund / Use-Case: Ein Kunde von mir möchte migrieren: von VMware + Veeam auf Proxmox VE + PBS und sucht nach einer sinnvollen Lösung für das Offline-Backup. Täglich soll via Copy-Job auf USB 9TB an Fullbacks transferiert werden, sodass auf dem USB-Datenträger ein vollwertiger Full-Zustand...

This is a additonal setting that can be set "NUMA node per L3 cache" but did you also try to set "Nume nodes per node" to a different value then the default? On my side its set to 1 at default. I am interested in knowing if more numa nodes is better or worse for performance. Edit: guess I need...

Should NUMA Nodes per Socket be set to 1/2/4 in BIOS? Whats the best configuration for a single socket amd-cpu? In VM Hardware I would enable NUMA. Or is it better to set NUMA-Nodes to 1 and not use numa in vm hardware? Has someone made some extensive tests or can answer this question?

Its visible when you install a new proxmox node with ceph, there you can select it already.

Seems like tentacle has landed in no-subscription now (noticed it today) :-)

Nothing has changed regarding this afaik. You need switchport with multiple vlans on it (vlan trunk), then create a vmbr0 with a bridged port (that connects to your switchport that(s) with the vlans on it). Then USE the sdn, as its the most simple way to do vlans for vm-access. If you also want...

Crimson is not included in pveceph yet.

I personally would look for journalctl -u pveproxy.service limiting with --since to the time-frame where you had that error. If it was 5 minutes ago for example it is: journalctl -u pveproxy.service --since "5 minutes ago"

Will ZFS 2.4 will come with a update within the 9.1.x window or is it planned for 9.2 (april?) https://git.proxmox.com/?p=zfsonlinux.git;a=summary

Yeah the customer thought it is needed and forgot about it. Setting it to 1 did the job :-)

multipath_component_detection = 0 needed to be set to 1 in /etc/lvm/lvm.conf - customer set the parameter, which caused this error. Thank you all for your help, I learned alot about iSCSI. Even created my own iSCSI-Target in a debian-vm to test it out myself (where it did work on first try)