initramfs-tools has a number of RC bugs open in jessie and more in unstable. I believe we will need freeze exceptions to address these.
The current versions are 0.116 in jessie, 0.118 in sid. #616689 Root-on-LVM setup fails often due to timing issues (and 9 merged bugs) ------------------------------------------------------------------------------ We may not wait long enough for the root device to appear, or depending on timing we may fail to run the necessary configuration commands. This is a longstanding issue but seems to have been exacerbated since wheezy by kernel changes. I don't think we can ignore this for jessie. It is at least partly the fault of other packages, and in any case is likely to require changes to hook scripts in some or all of cryptsetup, lvm2 and mdadm. #678696 includes a proposed fix but it is not complete. Ubuntu apparently has fixed this but their branch of initramfs-tools has diverged significantly. I am attempting to cherry-pick the necessary changes. I was promised some help with this by the Ubuntu maintainers. #652459 initramfs-tools: [patch] Please support mounting of /usr in the initramfs --------------------------------------------------------------------------------- This was opened with severity normal, but probably should be important or serious - lots of things break (usually subtly) if /usr is separate and is not mounted before the real init system starts. The bug was closed in 0.117, but only systemd is currently compatible with this behaviour so 0.118 made the behaviour conditional on init being systemd. Note that mounting /usr required that we also fsck it (#708000, fixed at the same time), as fsck programs generally refuse to operate on a mounted filesystem other than root. This brings its own issues (#765614 and #766448 listed below, and #767040 assigned to e2fsprogs). #762984 initramfs-tools: Alert! /dev/vg0/usr does not exist ----------------------------------------------------------- This is sort of a regression in 0.117, in that it relates to mounting /usr, but I think it's effectively a duplicate of #616689. #765614 Missing fsck.<type> should be non-fatal ----------------------------------------------- This is a regression in 0.117; I have a candidate fix but need to test it. #766448 root fstype "auto" in /etc/fstab guaranteed unbootable -------------------------------------------------------------- This is a regression in 0.117 and likely to be fixable fairly easily. #766920 initramfs-tools: update-initramfs makes system unbootable due to missing rootfs --------------------------------------------------------------------------------------- This was reported against 0.116 and I'm not convinced that it's a recent regression. It only affects odd lilo configurations, and I think that a change I made in 0.118 will change the behaviour from producing a broken initramfs to a fatal error at build time. Ben. -- Ben Hutchings The program is absolutely right; therefore, the computer must be wrong.
signature.asc
Description: This is a digitally signed message part
