Yes, please remove them from rc2.d to rc5.d, but keep the rcS.d link. So
now I understand why the boot went completely wrong, I just don't know
at all how these symlinks came into rc[2-5].d/ in the first place. For
this particular case we can probably just ship a dummy lvm2.service,
then we don't n
Yeah, for some reason my system has them in in *all* the rcX.d
directories. rc2-rc5 all have start scripts (S05lvm2) and the others
have kill scripts, except for rcS.d, which actually has nothing. I
guess I should delete all of them and add one in rcS.d? Should the one
in rcS.d be a start or kil
Ooh, something just occurred to me! sysv-generator doesn't actually read
the information from "Default-Start:", but from the links in /etc/rc*,
just like sysvinit would do. What's the output of
ls -l /etc/rc*/*lvm2*
? There should be one link in /etc/rcS.d, but *none* in /etc/rc2.d or
another n
Thanks. The init script works as expected, but the generated .service is
missing "DefaultDependencies=no". Unfortunately the sysv generator
doesn't have that many debug messages, so if you just run it with logs
that won't tell me much yet. I think I'll need to prepare one with extra
debugging infor
OK, here it is:
michael@mamarley-laptop:~$ cat /etc/init.d/lvm2
#!/bin/sh
### BEGIN INIT INFO
# Provides: lvm2 lvm
# Required-Start:mountdevsubfs
# Required-Stop:
# Should-Start: udev mdadm-raid cryptdisks-early multipath-tools-boot
# Should-Stop: umountroot mdadm-raid
# X-
> Requires=basic.target
OK, that's surprising, and the cause for the woes.
Can you please attach your /etc/init.d/lvm2? It might differ from the
default and add some more dependencies.
Can you please also give me the output of
systemctl show lvm2 | grep Path
cat /run/systemd/generator.late/
Sure, it says:
Requires=basic.target
Wants=system.slice
Before=multi-user.target shutdown.target graphical.target checkfs.service
local-fs.target
After=system.slice basic.target cryptdisks-early.service mdadm-raid.service
systemd-udevd.service mountdevsubfs.service systemd-journald.socket
multi
Ah, thanks for the journal. That shows that early boot is completely
messed up due to dependency cycles. This also affected systemd-tmpfiles-
setup.service:
Mar 12 05:56:32 mamarley-laptop systemd[1]: Found ordering cycle on
basic.target/start
Mar 12 05:56:32 mamarley-laptop systemd[1]: Found dep
** Summary changed:
- apparmor.service starts before remounting r/w, fails
+ ordering cycles with early boot images on lvm2.service
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1431107
Title:
orde
