Jesse,
On Sun, Sep 26, 2021 at 08:45:39PM -0300, Jesse Smith wrote:
> In each case insserv detected that the avahi-daemon script should be
> marked as K01 instead of K02. Which is good, that is consistent with
> what I have on my system and it looks to be correct. Once this change
> was made none of the three versions of insserv suggested switching
> avahi-daemon back to K02. ie On my system insserv does not perform the
> "toggle" action when given the uploaded init scripts.
>
> Which means when any recent version of insserv is run against just the
> init.d scripts provided it sets avahi-daemon to be K01 and leaves it
> that way, future runs don't toggle the symlink back. At least that's
> been my finding.
>
> Can anyone else confirm that running "insserv -p etc-stripped/init.d" on
> the uploaded scripts always sets avahi-daemon to K01, or do you get the
> "toggle" behaviour in the bug report?
Not quite.
Testing both versions 1.18.0-2 and 1.23.0-1 (with the same results), I find that
initially insserv made no changes, leaving the links at K02. If the links are
deleted they are recreated as K01.
I can't reproduce any toggle behaviour. However, if I change
rc0.d/K01avahi-daemon to rc0.d/K02avahi-daemon, insserv changes the links in
rc{1,6}.d to K02. That surprised me as I rather expected the link in rc0.d to be
changed back to K01.
This also happens in reverse if all the links are K02 and rc0.d is changed to
K01, the next insserv invocation changes all the links to K01.
So it seems as if insserv chooses K01 by default, but doesn't change K02 to
K01. Furthermore, higher numbers are reduced to K02, but never K01.
> If others see the same behaviour I do, then I'm guessing there is
> something else on the reporting system with the error which causes a
> conflict. But if other people are seeing the "toggle" between K01 and
> K02 on the provided scripts then I'm guessing things are just working
> correctly for me due to an oversight or quirk of my system. Any feedback
> on confirming or disputing my findings is appreciated.
Thorsten, I am wondering if you have anything in /etc/insserv/overrides or
changes to /etc/insserv.conf*?
Mark
