[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#728486: Current patch for resolving lvm/systemd compatibility



Hi Bastian,

Bastian Blank <waldi@debian.org> writes:
> On Sat, Jan 18, 2014 at 11:47:11AM +0100, Michael Stapelberg wrote:
>> --- /dev/null
>> +++ b/debian/lvm2-activation-early.service
>
> Wrong name.
Renaming them to lvm2{,-early}.service as you suggested is fine with me.

>> +[Unit]
>> +Description=Activation of LVM2 logical volumes
>> +Documentation=man:lvm(8) man:vgchange(8)
>> +DefaultDependencies=no
>> +After=systemd-udev-settle.service
>> +Before=cryptsetup.target
>
> This have to be "cryptdisks.service"
No. cryptdisks{,-early}.service are actually masked in the systemd
package. cryptsetup.target is the correct unit to refer to here, see
also systemd.special(7):

cryptsetup.target
A target that pulls in setup services for all encrypted block devices.

Units called e.g. systemd-cryptsetup@sda2_crypt.service are dynamically
generated from /etc/crypttab using a systemd generator and will then be
part of cryptsetup.target.

>> +usr/lib/tmpfiles.d/lvm2.conf
>
> Undocumented in the changelog.
The changelog was merely an example, so feel free to phrase that as you
want. In case you want some input from me, I’d call it “Install
tmpfiles.d(5) configuration for systemd”.

As for your other mail which suggests shipping symlinks to enable the
unit files, are you sure that there is absolutely no use-case in which
users might want to disable the service files? Having the symlinks in
the package will prevent that entirely.

Also, why do you want lvm2-early.service hooked into cryptsetup.target
instead of local-fs.target, as upstream suggests?

-- 
Best regards,
Michael


Reply to: