aenix-io
mount tmpfs: operation not permitted
I'm on a VPS of the cloud provider OVHCloud OS: Ubuntu 24.04
ubuntu@vps-a79e4302:~$ boot-to-talos Target disk [/dev/sda]: Talos installer image [ghcr.io/cozystack/cozystack/talos:v1.10.5]: Add networking configuration? [yes]: yes Interface [enp0s3]: IP address [aa.aa.aaa.aaa]: Netmask [255.255.255.255]: Gateway (or 'none') [bb.bb.bb.b]: Configure serial console? (or 'no') [ttyS0]:
Summary: Image: ghcr.io/cozystack/cozystack/talos:v1.10.5 Disk: /dev/sda Extra kernel args: ip=aa.aa.aa.aaa::bb.bb.bb.b:255.255.255.255::enp0s3::::: console=ttyS0
WARNING: ALL DATA ON /dev/sda WILL BE ERASED!
Continue? [yes]: yes
2025/10/17 15:19:12 created temporary directory /tmp/installer-4012601756 2025/10/17 15:19:12 mount tmpfs: operation not permitted
What to do to make it work?
The error means your VPS environment doesn't allow mounting tmpfs, which is required for boot-to-talos to work. This restriction is common on container-based VPS platforms (like LXC or OpenVZ), but not on fully virtualized (KVM) servers. There is no workaround in boot-to-talos for this limitation—mounting tmpfs is mandatory for the installer to proceed, and the code doesn't provide an alternative flow or privilege escalation for restricted environments.
To resolve this, check if your VPS is container-based (for example, run systemd-detect-virt or virt-what). If it reports LXC, OpenVZ, or similar, you will not be able to use boot-to-talos on this VPS. If possible, switch to a KVM-based VPS or a dedicated server, which will allow tmpfs mounts and let the installer run as intended. If you are already on a KVM VPS and still see this error, make sure you are running the installer as root (with sudo).
Reference: boot-to-talos requires tmpfs mount and will fail in restricted/containerized VPS environments
To reply, just mention @dosu.
How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other 
![dosubot[bot] avatar](https://avatars.githubusercontent.com/in/324583?v=4 "Published by a pub.dev verified publisher"){kind=small}
@dosu
Luckily it's a KVM-based VPS:
ubuntu@vps-a79e4302:~$ systemd-detect-virt
kvm
But, I tried to run the installer as root and got "command not found" :
ubuntu@vps-a79e4302:~$ sudo boot-to-talos
sudo: boot-to-talos: command not found
If I run the installer as non-root user it finds the command but the mount operation is not permitted:
ubuntu@vps-a79e4302:~$ boot-to-talos
Target disk [/dev/sda]:
This happens because boot-to-talos is installed in a user-local directory (like ~/.local/bin), which is in your PATH as a regular user but not when using sudo. To fix this, run which boot-to-talos to see where it’s installed, then use the full path with sudo, for example: sudo ~/.local/bin/boot-to-talos.
Alternatively, you can move or reinstall boot-to-talos to /usr/local/bin, which is in root’s PATH, so sudo boot-to-talos will work as expected. This is a common situation on Ubuntu and similar systems when installing user-local binaries.
Reference: install script logic
To reply, just mention @dosu.
How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other
@dosu
After moving boot-to-talos to /usr/local/bin it seems working with sudo
But copying the image.raw to /dev/sda seems taking too long: already more than 35 minutes...
ubuntu@vps-a79e4302:~$ sudo mv .local/bin/boot-to-talos /usr/local/bin/
ubuntu@vps-a79e4302:~$ sudo boot-to-talos
Target disk [/dev/sda]:
Talos installer image [ghcr.io/cozystack/cozystack/talos:v1.10.5]:
Add networking configuration? [yes]:
Interface [enp0s3]:
IP address [37.59.118.201]:
Netmask [255.255.255.255]:
Gateway (or 'none') [37.59.96.1]:
Configure serial console? (or 'no') [ttyS0]:
Summary:
Image: ghcr.io/cozystack/cozystack/talos:v1.10.5
Disk: /dev/sda
Extra kernel args: ip=37.59.118.201::37.59.96.1:255.255.255.255::enp0s3::::: console=ttyS0
WARNING: ALL DATA ON /dev/sda WILL BE ERASED!
Continue? [yes]: yes
2025/10/17 16:05:54 created temporary directory /tmp/installer-2866190222
2025/10/17 16:05:54 pulling image ghcr.io/cozystack/cozystack/talos:v1.10.5
2025/10/17 16:05:54 No proxy configured, using direct connection
2025/10/17 16:05:55 extracting image layers
2025/10/17 16:05:55 No proxy configured, using direct connection
2025/10/17 16:05:59 creating raw disk /tmp/installer-2866190222/image.raw (2 GiB)
2025/10/17 16:05:59 attached /tmp/installer-2866190222/image.raw to /dev/loop0
2025/10/17 16:05:59 starting Talos installer
2025/10/17 16:05:59 running Talos installer v1.10.5
2025/10/17 16:05:59 WARNING: config validation:
2025/10/17 16:05:59 use "worker" instead of "" for machine type
2025/10/17 16:05:59 created EFI (C12A7328-F81F-11D2-BA4B-00A0C93EC93B) size 104857600 bytes
2025/10/17 16:05:59 created BIOS (21686148-6449-6E6F-744E-656564454649) size 1048576 bytes
2025/10/17 16:05:59 created BOOT (0FC63DAF-8483-4772-8E79-3D69D8477DE4) size 1048576000 bytes
2025/10/17 16:05:59 created META (0FC63DAF-8483-4772-8E79-3D69D8477DE4) size 1048576 bytes
2025/10/17 16:05:59 formatting the partition "/dev/loop0p1" as "vfat" with label "EFI"
2025/10/17 16:05:59 formatting the partition "/dev/loop0p2" as "zeroes" with label "BIOS"
2025/10/17 16:05:59 formatting the partition "/dev/loop0p3" as "xfs" with label "BOOT"
2025/10/17 16:05:59 formatting the partition "/dev/loop0p4" as "zeroes" with label "META"
2025/10/17 16:05:59 copying from io reader to /boot/A/vmlinuz
2025/10/17 16:05:59 copying from io reader to /boot/A/initramfs.xz
2025/10/17 16:05:59 writing /boot/grub/grub.cfg to disk
2025/10/17 16:05:59 executing: grub-install --boot-directory=/boot --removable --efi-directory=/boot/EFI /dev/loop0
2025/10/17 16:05:59 installation of v1.10.5 complete
2025/10/17 16:06:00 Talos installer finished successfully
2025/10/17 16:06:00 remounting all filesystems read-only
2025/10/17 16:06:00 copy /tmp/installer-2866190222/image.raw → /dev/sda
Is that fine and normal?
It's not normal for copying a 2 GiB image to /dev/sda to take over 35 minutes on most VPS setups. In boot-to-talos, this step uses a synchronous write approach—each 4 MiB chunk is written and then flushed to disk with fsync, which can be very slow if your VPS disk performance is limited or heavily contended. There's no progress feedback during this step, so it may look stalled even if it's just slow source.
This is usually caused by the underlying storage backend of your VPS, not by a bug in boot-to-talos. You can check your disk write speed with a command like:
dd if=/dev/zero of=testfile bs=4M count=512 conv=fdatasync
If you see speeds below 10–20 MB/s, that's likely the cause. If performance is unacceptably slow, you may want to contact your provider or try a different VPS plan.
To reply, just mention @dosu.
How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other
Hey @raphael10-collab it seems boot-to-talos method does not work well with your setup.
However you can also try kexec method to install Talos Linux, here is a doc:
- https://cozystack.io/docs/install/talos/kexec/#1-install-kexec-tools-package
