+++ This bug was initially created as a clone of Bug #5159 +++ Description of problem: Hi! I'm doing minimal installation (without X Window), and I got when halting error that are visible on the attachement My partitions sda1 /boot sda2 encrypted LVM * / * swap * /home Steps to Reproduce: 1. network installation (nonfree iso) 2. see errors at halt time
Created attachment 1893 [details] dm errors
Reproduced here in my VM. Passing acpi=off on the command line makes the messages vaguely readable too :D
Status: NEW => ASSIGNED
Assignee: bugsquad => mageia
Should be fixed in latest dracut. Systemd will still output these messages, but it is impossible for systemd to do more here anyway (it could in theory be a little cleverer about how to tear down DM's such that it won't even try to tear down the DM if the rootfs is somewhere within it, but that's a mostly cosmetic issue really). The difference now is that dracut will take over again, and do it's own tidy up of stuff. Sadly after all this, it still seems that dracut doesn't actually shutdown the LVM or crypt anyway (I cannot see any shutdown hooks to call vgchange -an) but this is an upstream problem. We can't realistically do much more than this (other than by patching dracut, but I'm sure this is on the roadmap anyway and this is already better than we were able to do before with sysvinit and mkinitrd so I thing that's an improvement. Therefore, closing.
Status: ASSIGNED => RESOLVEDResolution: (none) => FIXED
(In reply to comment #3) > Should be fixed in latest dracut. Systemd will still output these messages, but > it is impossible for systemd to do more here anyway (it could in theory be a > little cleverer about how to tear down DM's such that it won't even try to tear > down the DM if the rootfs is somewhere within it, but that's a mostly cosmetic > issue really). > > The difference now is that dracut will take over again, and do it's own tidy up > of stuff. > > Sadly after all this, it still seems that dracut doesn't actually shutdown the > LVM or crypt anyway (I cannot see any shutdown hooks to call vgchange -an) but > this is an upstream problem. We can't realistically do much more than this > (other than by patching dracut, but I'm sure this is on the roadmap anyway and > this is already better than we were able to do before with sysvinit and > mkinitrd so I thing that's an improvement. > > Therefore, closing. Thank you for fixing it. Please report it upstream.