[SOLVED] Are linux headers required or save to delete? Even if they are old
- Thread starter diversity
- Start date
Update your Proxmox: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#system_software_updates
After enabling the right repositories: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#sysadmin_package_repositories
But sometimes you need to uninstall/remove old kernel versions manually. And you need to reboot (to run new kernel versions) sometimes.
After enabling the right repositories: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#sysadmin_package_repositories
But sometimes you need to uninstall/remove old kernel versions manually. And you need to reboot (to run new kernel versions) sometimes.
The package manager typically recommends the versions you need based on what you're currently running or asked it to do. Why you're currently running what you're running is another question, as mentioned above, could be repo's not set up properly, perhaps you have the kernel pinned for a reason or a dependency, or you upgraded from Proxmox 7->8 and something went weird?
You can do apt autoremove (after updating and rebooting) to uninstall older kernels you no longer need, but it is useful to keep a few versions of the kernel around in case there are issues with the current version and definitely do not force-remove them. Keeping the packages on your system wastes a few MB of disk space, nothing more, that's different than what is currently "running" on your system. It would be weird to have eg. CUPS running on your Proxmox server, it certainly does not by default.
As far as your specific case, are you using CUDA/nVIDIA drivers on the hypervisor side? That MAY be why the kernel is stuck, if you installed a specific version of the driver, it needs a specific version of the kernel - there are meta-packages that will update every version (which may be a pain in itself), but if you specified eg. 525 series of drivers, they are going to be the 525 drivers as you asked.
You shouldn't need (the public) CUDA/drivers on the Proxmox server though, if you do for some odd reason, you may want to keep the old kernel version while you test whether jumping to 6.8 works, you will need to upgrade the driver (if that works) or if it doesn't (if the leap is too great) uninstall the packages, disable the repo, make sure everything gets up-to-date, then re-install from there (you will need 'current' versions of the commercial drivers (550) for vGPU to work on 6.8, see my other thread).
You can do apt autoremove (after updating and rebooting) to uninstall older kernels you no longer need, but it is useful to keep a few versions of the kernel around in case there are issues with the current version and definitely do not force-remove them. Keeping the packages on your system wastes a few MB of disk space, nothing more, that's different than what is currently "running" on your system. It would be weird to have eg. CUPS running on your Proxmox server, it certainly does not by default.
As far as your specific case, are you using CUDA/nVIDIA drivers on the hypervisor side? That MAY be why the kernel is stuck, if you installed a specific version of the driver, it needs a specific version of the kernel - there are meta-packages that will update every version (which may be a pain in itself), but if you specified eg. 525 series of drivers, they are going to be the 525 drivers as you asked.
You shouldn't need (the public) CUDA/drivers on the Proxmox server though, if you do for some odd reason, you may want to keep the old kernel version while you test whether jumping to 6.8 works, you will need to upgrade the driver (if that works) or if it doesn't (if the leap is too great) uninstall the packages, disable the repo, make sure everything gets up-to-date, then re-install from there (you will need 'current' versions of the commercial drivers (550) for vGPU to work on 6.8, see my other thread).
Last edited: