Bug#754404: /usr/sbin/invoke-rc.d: line 288: /sbin/runlevel: No such file or directory

[Simon McVittie]

severity 754404 minor
thanks

On 10/07/14 21:51, Michael Biebl quoted:
>> Processing triggers for dbus (1.8.6-1) ...
>> /usr/sbin/invoke-rc.d: line 288: /sbin/runlevel: No such file or directory
>> [....] system message bus already started; not starting.?25l1G[32m ok 39;49m?12l?25h.

For what it's worth, this is unnecessary in almost all cases -
restarting "dbus-daemon --system" other than by rebooting is not a
supported action, so the maintainer script just starts it if it isn't
running, and doesn't restart it if it is already running. The failure to
run /sbin/runlevel appears to have been harmless: invoke-rc.d still
correctly determines that you already have a "dbus-daemon --system" running.

Forcing configuration to be reloaded (which is the part of dbus.postinst
that's actually relevant to the trigger) is done via a D-Bus call, not
via the init script, so that it can be the same code path regardless of
whether we booted with sysvinit, systemd or Upstart; so we're not
relying on invoke-rc.d for that either.

You could argue that the "make sure it has started" boilerplate
shouldn't be run in response to a trigger, only in response to new
installation. That code is provided by dh_installinit, so the dbus
package has no control over it, other than using --noscripts and
supplying equivalent code itself (which I'd rather avoid if possible).

    S