On Mon, Sep 05, 2011 at 08:30:35PM +0200, martin f krafft wrote: > severity 640541 wishlist > retitle 640541 update initrd for all kernels, or document that this needs to > be done manually > thanks > > also sprach Paul Menzel <pm.deb...@googlemail.com> [2011.09.05.1933 +0200]: > > I had `mdadm` on hold for quite a while and updated the package today. > > > > [UPGRADE] mdadm 3.1.1-1 -> 3.1.4-1+8efb9d1 > > > > The confusing part is that the initramfs was upgraded only for my > > current self-build Linux kernel and not for all the other Linux kernels. > > rhiswas a decision we purposely took way back when, because we > wanted to prevent mdadm from breaking even other known-working boot > configurations. Surely, that is not an ideal solution. > > So we should either update all initrds at all times, or document > that you need to do that. > > I think that there is some sort of initramfs policy against the > former. Max, can you shed some light onto that?
Yes indeed the policy is to be in the conservative side and only update the newest initramfs and thus not potentially break the boot of boxes. The local admin can easily overrule this in update-initramfs.conf You can close the bug as it's shortly documented in man update-initramfs.conf. Also for slower boxes rebuilding all initramfs would slow down updates considerably. The second avoided trouble is full disks. happy hacking maks -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
