As the system requirements for Proxmox Datacenter Manager are so low for non-commercial use, are there ay plans to make an ARM compatible version? I know the project is still in it's infancy but it would be really cool to have this running on a pi for example. Currently I have this running in an LXC container in my Proxmox cluster which kind of defeats the point
Proxmox Datacenter Manager on ARM?
- Thread starter robj18
- Start date
It definitely has a right to be there, but how useful it really is is another matter.
Maybe it's something that will happen in the future Development.
You could Open a Bug / Feature-request in BugZilla.
Maybe it's something that will happen in the future Development.
You could Open a Bug / Feature-request in BugZilla.
Honestly, I think it's an incredibly important step....
There's legit benefit to having arm proxmox hosts....
I did a quick test.... geekbench scores on a raspberry pi4 ran circles around the same tests running in an emulated arm64 qemu VM running on xeon hardware with gobs more IO/CPU/Mem...
I want to be able to have, for example, GitLab CI ARM64 runners for doing container builds .... emulating arm64 in docker to do multiarch builds is really painfully inefficient ...
havin the ability to register some ARM64 proxmox nodes as workload trucks could be really useful but getting to that point isn't gonna happen quickly... coz lots of moving parts .... but having the datacenter manager component able to run on a Pi, for example, would be (in my opinion) a fantastic way to start getting the core components of pxm working while sandboxing multiarch interop problems as a problem for future-self .....
There's legit benefit to having arm proxmox hosts....
I did a quick test.... geekbench scores on a raspberry pi4 ran circles around the same tests running in an emulated arm64 qemu VM running on xeon hardware with gobs more IO/CPU/Mem...
I want to be able to have, for example, GitLab CI ARM64 runners for doing container builds .... emulating arm64 in docker to do multiarch builds is really painfully inefficient ...
havin the ability to register some ARM64 proxmox nodes as workload trucks could be really useful but getting to that point isn't gonna happen quickly... coz lots of moving parts .... but having the datacenter manager component able to run on a Pi, for example, would be (in my opinion) a fantastic way to start getting the core components of pxm working while sandboxing multiarch interop problems as a problem for future-self .....
Actually, if you search for "PiMox", you can find the install howto for proxmox (special, arm64 repo) onto Debian 12 on Raspberry Pi 4/5 and CM4/5.
I am using a hybrid cluster of 1x CM5 (main node), 1x RPi4 (using usb ssd!) and 1x Intel Xeon old server (as the only AMD64/x86_64 node).
Of course, migrating between different arch's isn't available, but I can easily start an x86_64/Intel/AMD arch vm on the Pi's, as the correct debs are available on PiMox/Deb12 on ARM64 -> of course, this latter approach is inefficient, I rather use the Intel-based server for those VMs.
Currently, LXCs are native-arch only (meaning: I can only start an Intel-arch LXC on Intel, and ARM64/ARM32 arch on the Pi's).
Hope, these help!
pimox is an unofficial community product though so not suited for usage in corporate environments (e.G. for using ProxmoxVE on a ARM64 cloud instance)
This is unlikely to change since LXCs always run in the kernel space of the host, so the applications running inside the LXC needs to be compiled for the host architecture. Now in theory you could use the qemu to run non-native binaries (e.G. x86 on arm64 and vice verca) in an emulator inside the lxc but if you do this you could also setup a VM for that (because ProxmoxVE VMs are basically also just running inside qemu virtual machines)
Last edited:
Unofficial: sure, I forgot to mention it. No official support from Proxmox either, of course! However, I hope, Proxmox will add ARM64 as an official arch. If not for Proxmox VE (at first), then for Datacenter and PBS should be first...
LXC & kernel arch: sure, kernel-level is the main advantage of the LXC subsystem, so no problem there, for me.