I found this portion in
However, I'm using systemd-boot and the default EFI partition is 512MB on most of the machines in my clusters, so when I perform an upgrade, it sometimes fails because the ESP runs out of space when apt tries to keep four or more kernels on there. Or it would have succeeded but fails because the new kernel is installed before the old is cleaned up.
Yes, I did manually convert from grub to systemd-boot using proxmox-boot-tool for ....reasons. And I found that if you put rootfs directly on ZFS the installer defaults to 1GB ESP using systemd-boot and the problem would not be there, but these systems have been upgraded from PVE 5-6-7 already and it would be a hassle to reinstall them all, or have to revert to grub...
It would be great if we could shortcut this logic somewhat. For myself, it would be sufficient to keep
down to
which does what I described (maybe?)
And I believe it is okay to remove
/usr/share/pve-kernel-helper/scripts/functions
Code:
# Mark as not-for-autoremoval those kernel packages that are:
# - the currently booted version, if still installed
# - the kernel version we've been called for
# - the latest kernel version (as determined by debian version number)
# - the second-latest kernel version
# - the latest kernel version of each series (e.g. 4.13, 4.15, 5.0) by
# marking the meta-packages
# - the currently pinned kernel if anyHowever, I'm using systemd-boot and the default EFI partition is 512MB on most of the machines in my clusters, so when I perform an upgrade, it sometimes fails because the ESP runs out of space when apt tries to keep four or more kernels on there. Or it would have succeeded but fails because the new kernel is installed before the old is cleaned up.
Yes, I did manually convert from grub to systemd-boot using proxmox-boot-tool for ....reasons. And I found that if you put rootfs directly on ZFS the installer defaults to 1GB ESP using systemd-boot and the problem would not be there, but these systems have been upgraded from PVE 5-6-7 already and it would be a hassle to reinstall them all, or have to revert to grub...
It would be great if we could shortcut this logic somewhat. For myself, it would be sufficient to keep
- The currently running kernel
- The latest kernel from the metapackage required by the
proxmox-vepackage
/usr/share/pve-kernel-helper/scripts/functions:
Code:
kernels="$(cat <<-EOF
$running_version
$install_version
$manual_kernels
$latest_2_versions
$series_metapackages
$oldseries_latest_kernel
$pinned_kernel
$nextboot_kernel
EOF
)"down to
Code:
kernels="$(cat <<-EOF
$running_version
$install_version
$manual_kernels
$pinned_kernel
$nextboot_kernel
EOF
)"which does what I described (maybe?)
And I believe it is okay to remove
$series_metapackages in this case too since proxmox-ve depends on the last metapackage anyway?
Last edited:
