In the initramfs phase LVM is not activated and I am being dropped to rescue shell. If I manually run lvm_scan and then exit from the shell then dracut continues to boot. All LVM PVs are on mdraid arrays. This is not the same issue as in bug 432576. I have no problems with dracut-022-r5. And I have systems which is not suffer from this issue. On this systems, there is no mdraid. $ emerge -pv dracut These are the packages that would be merged, in order: Calculating dependencies... done! [ebuild R ] sys-kernel/dracut-023-r1 USE="device-mapper -debug -net -optimization (-selinux)" DRACUT_MODULES="lvm mdraid -biosdevname -bootchart -btrfs -caps -cifs -crypt -crypt-gpg -crypt-loop -dmraid -dmsquash-live -gensplash -iscsi -livenet -multipath -nbd -nfs -plymouth -ssh-client -syslog -systemd" 0 kB Total: 1 package (1 reinstall), Size of downloads: 0 kB Reproducible: Always
Created attachment 326066 [details] init.log rd.debug added to kernel cmdline
Created attachment 326068 [details] dracut-023.log dracut -L 6 output
*** This bug has been marked as a duplicate of bug 432576 ***
(In reply to comment #3) > > *** This bug has been marked as a duplicate of bug 432576 *** I think this is not a duplicate. Bug 432576: lvm is not activated with both dracut-022-r5 and 023. Reporter doesn't use mdraid. This bug: lvm is not activated only with dracut-023 and only on a system with mdraid.
OK. I've sorted out this problem. dracut-23 should depend on >=sys-fs/udev-174. According to udev's changelog "path_id" builtin appeared in udev-174 and "blkid" builtin appeared in udev-176. http://upstream-tracker.org/changelogs/libudev/181/changelog.html After replacing these lines: $ egrep 'blkid|path_id' /usr/lib64/dracut/modules.d/95udev-rules/*-persistent-storage.rules.ORIG /usr/lib64/dracut/modules.d/95udev-rules/59-persistent-storage.rules.ORIG:KERNEL=="cciss[0-9]*", IMPORT{builtin}="blkid" /usr/lib64/dracut/modules.d/95udev-rules/59-persistent-storage.rules.ORIG:KERNEL=="nbd[0-9]*", IMPORT{builtin}="blkid" /usr/lib64/dracut/modules.d/95udev-rules/59-persistent-storage.rules.ORIG:IMPORT{builtin}="blkid" /usr/lib64/dracut/modules.d/95udev-rules/59-persistent-storage.rules.ORIG:IMPORT{builtin}="blkid" /usr/lib64/dracut/modules.d/95udev-rules/61-persistent-storage.rules.ORIG:ENV{DEVTYPE}=="disk", ENV{ID_PATH}=="", DEVPATH!="*/virtual/*", IMPORT{builtin}="path_id" on these: $ egrep 'blkid|path_id' /usr/lib64/dracut/modules.d/95udev-rules/*-persistent-storage.rules /usr/lib64/dracut/modules.d/95udev-rules/59-persistent-storage.rules:KERNEL=="cciss[0-9]*", IMPORT{program}="/sbin/blkid -o udev -p $tempnode" /usr/lib64/dracut/modules.d/95udev-rules/59-persistent-storage.rules:KERNEL=="nbd[0-9]*", IMPORT{program}="/sbin/blkid -o udev -p $tempnode" /usr/lib64/dracut/modules.d/95udev-rules/59-persistent-storage.rules:IMPORT{program}="/sbin/blkid -o udev -p $tempnode" /usr/lib64/dracut/modules.d/95udev-rules/59-persistent-storage.rules:IMPORT{program}="/sbin/blkid -o udev -p $tempnode" /usr/lib64/dracut/modules.d/95udev-rules/61-persistent-storage.rules:ENV{DEVTYPE}=="disk", ENV{ID_PATH}=="", DEVPATH!="*/virtual/*", IMPORT{program}="path_id %p" LVM is activated now!
I'm using udev-171
(In reply to comment #5) > OK. I've sorted out this problem. dracut-23 should depend on > >=sys-fs/udev-174. According to udev's changelog "path_id" builtin appeared > in udev-174 and "blkid" builtin appeared in udev-176. Sorry, >=sys-fs/udev-176 of course
Same issue with dracut-024. Of course I don't forget to add "rd.auto" to kernel cmdline.
Created attachment 327828 [details, diff] dracut-024-udev-builtins.patch Old udev compatibility patch. Works for me, but it seems ugly. :) Don't forget to `chmod +x modules.d/95udev-rules/builtins-check.sh`
Created attachment 327890 [details, diff] dracut-024-udev-builtins-v2.patch Another patch using different approach.
Upstream hasn't changed
(In reply to comment #11) > Upstream hasn't changed ...send to quickly. :-) Upstream hasn't changed dependency on >=udev-166 to 176, so let's try first to use your patches, Alexander.
Thank you for the report and the patch! Boots for me with latest udev. If you could retest dracut-024-r1 with any stable udev post 166, I'd be grateful. :-) *dracut-024-r1 (01 Nov 2012) *dracut-023-r3 (01 Nov 2012) 01 Nov 2012; Amadeusz Żołnowski <aidecoe@gentoo.org> +files/024-0001-Fallback-to-external-blkid-and-path_id.patch, +files/023-0003-Fallback-to-external-blkid-and-path_id.patch, +dracut-023-r3.ebuild, +dracut-024-r1.ebuild: Fixes bug #437700. Thanks to Alexander Tsoy <alexander@tsoy.me> for the report and the patch. Also bumped dhcp version dependency to >=4.2.4_p2-r1 wrt bug #437108.
FYI: I have pushed your patch upstream.
(In reply to comment #13) > Thank you for the report and the patch! > export UDEVVERSION is not needed in init.sh, so you can drop this part from the patch
(In reply to comment #15) > (In reply to comment #13) > > Thank you for the report and the patch! > > > > export UDEVVERSION is not needed in init.sh, so you can drop this part from > the patch And you forgot to update ebuilds. epatch is missing and maybe chmod is necessary.
Created attachment 327926 [details, diff] 024-0001-Fallback-to-external-blkid-and-path_id.patch
files/024-0001-Fallback-to-external-blkid-and-path_id.patch is not used because: $ grep epatch -c sys-kernel/dracut/dracut-024-r1.ebuild 0
(In reply to comment #16) > ... and maybe chmod is necessary. I'm wrong. Executable bit is not necessary for hooks since they are simply sourced from init. :)
Sorry, here goes the fix. +*dracut-024-r2 (01 Nov 2012) +*dracut-023-r4 (01 Nov 2012) + + 01 Nov 2012; Amadeusz Żołnowski <aidecoe@gentoo.org> + files/024-0001-Fallback-to-external-blkid-and-path_id.patch, + files/023-0003-Fallback-to-external-blkid-and-path_id.patch, + -dracut-023-r3.ebuild, +dracut-023-r4.ebuild, -dracut-024-r1.ebuild, + +dracut-024-r2.ebuild: + Added missing epatch and improved patches a bit btw. +
Works for me with udev-171. Thank you, Amadeusz!
Thank you, too. :-)