Bug#989284: insserv: toggles rc0.d/{K02avahi-daemon => K01avahi-daemon} with every upgradel

[Mark Hindley]

Thorsten,

Thanks for this and sorry for taking some time to actually get round to trying
it.

On Wed, Oct 06, 2021 at 08:16:40PM +0200, Thorsten Glaser wrote:
> On Wed, 6 Oct 2021, Thorsten Glaser wrote:
> 
> > So I can verify this behaviour in an otherwise clean chroot.
> 
> And https://mops.tarent.de/.tmp/base.cow-bullseye-amd64.tar.xz is the
> chroot, just in case it is something about that as well.

I am afraid I am still unable to reproduce this behaviour in my own or your
supplied chroot.

The only thing that happens is a reordering of the .depend* files on the first
insserv invocation.  After that, there are no changes on subsequent runs.

mark at apollo:/var/cache/pbuilder% sudo tar xJf ~/Downloads/base.cow-bullseye-amd64.tar.xz
mark at apollo:~% sudo cp /tmp/etc-stripped.tgz /var/cache/pbuilder/build/cow.23212/tmp/
mark at apollo:~% cdiff --stat  /var/cache/pbuilder/build/cow.23212/tmp/etc-stripped{,1}
 .../init.d/.depend.boot                            | 20 ++++++++---------
 .../init.d/.depend.start                           | 26 +++++++++++-----------
 .../init.d/.depend.stop                            | 14 ++++++------
 3 files changed, 30 insertions(+), 30 deletions(-)
mark at apollo:~% cdiff --stat  /var/cache/pbuilder/build/cow.23212/tmp/etc-stripped{1,2}
 .../init.d/.depend.boot                            | 20 ++++++++---------
 .../init.d/.depend.start                           | 26 +++++++++++-----------
 .../init.d/.depend.stop                            | 14 ++++++------
 3 files changed, 30 insertions(+), 30 deletions(-)
mark at apollo:~% cdiff --stat  /var/cache/pbuilder/build/cow.23212/tmp/etc-stripped{1,3}
 .../init.d/.depend.boot                            | 20 ++++++++---------
 .../init.d/.depend.start                           | 26 +++++++++++-----------
 .../init.d/.depend.stop                            | 14 ++++++------
 3 files changed, 30 insertions(+), 30 deletions(-)
mark at apollo:~% cdiff --stat  /var/cache/pbuilder/build/cow.23212/tmp/etc-stripped{1,4}
 .../init.d/.depend.boot                            | 20 ++++++++---------
 .../init.d/.depend.start                           | 26 +++++++++++-----------
 .../init.d/.depend.stop                            | 14 ++++++------
 3 files changed, 30 insertions(+), 30 deletions(-)
mark at apollo:~% cdiff --stat /var/cache/pbuilder/build/cow.23212/tmp/etc-stripped{2,3}
mark at apollo:~% cdiff --stat /var/cache/pbuilder/build/cow.23212/tmp/etc-stripped{2,4}
mark at apollo:~% cdiff --stat -u /var/cache/pbuilder/build/cow.23212/tmp/etc-stripped{1,2}
 .../init.d/.depend.boot                            | 20 ++++++++---------
 .../init.d/.depend.start                           | 26 +++++++++++-----------
 .../init.d/.depend.stop                            | 14 ++++++------
 3 files changed, 30 insertions(+), 30 deletions(-)
                 
diff --git var/cache/pbuilder/build/cow.23212/tmp/etc-stripped1/init.d/.depend.boot var/cache/pbuilder/build/cow.23212/tmp/etc-stripped2/i
nit.d/.depend.boot                                                                                                                       
index 3fc1825..54d3471 100644
--- var/cache/pbuilder/build/cow.23212/tmp/etc-stripped1/init.d/.depend.boot
+++ var/cache/pbuilder/build/cow.23212/tmp/etc-stripped2/init.d/.depend.boot
@@ -1,4 +1,4 @@  
-TARGETS = mountkernfs.sh udev mountdevsubfs.sh keyboard-setup.sh bootlogd hwclock.sh hostname.sh checkroot.sh cryptdisks-early cryptdisks
 networking mountnfs.sh mountnfs-bootclean.sh mountall.sh mountall-bootclean.sh lvm2 x11-common early-rng-init-tools urandom brightness al
sa-utils checkfs.sh checkroot-bootclean.sh screen-cleanup stop-bootlogd-single bootmisc.sh mount-configfs kmod procps lm-sensors         
+TARGETS = mountkernfs.sh udev mountdevsubfs.sh keyboard-setup.sh bootlogd hwclock.sh hostname.sh checkroot.sh cryptdisks-early cryptdisks
 mountnfs.sh mountnfs-bootclean.sh mountall.sh mountall-bootclean.sh brightness networking lvm2 checkfs.sh urandom x11-common early-rng-in
it-tools alsa-utils mount-configfs kmod bootmisc.sh procps screen-cleanup stop-bootlogd-single lm-sensors checkroot-bootclean.sh         
 INTERACTIVE = udev keyboard-setup.sh checkroot.sh cryptdisks-early cryptdisks checkfs.sh
 udev: mountkernfs.sh
 mountdevsubfs.sh: udev
@@ -9,23 +9,23 @@ hostname.sh: bootlogd
 checkroot.sh: hostname.sh keyboard-setup.sh
 cryptdisks-early: checkroot.sh udev
 cryptdisks: cryptdisks-early lvm2
-networking: mountkernfs.sh mountall.sh mountall-bootclean.sh urandom procps
 mountnfs.sh: mountall.sh mountall-bootclean.sh networking
 mountnfs-bootclean.sh: mountall.sh mountall-bootclean.sh mountnfs.sh
-mountall.sh: checkfs.sh checkroot-bootclean.sh
+mountall.sh: lvm2 checkfs.sh checkroot-bootclean.sh
 mountall-bootclean.sh: mountall.sh
+brightness: mountall.sh mountall-bootclean.sh
+networking: mountkernfs.sh mountall.sh mountall-bootclean.sh urandom procps
 lvm2: cryptdisks-early bootlogd
+checkfs.sh: checkroot.sh cryptdisks lvm2
+urandom: hwclock.sh mountall.sh mountall-bootclean.sh
 x11-common: mountnfs.sh mountnfs-bootclean.sh
 early-rng-init-tools: hwclock.sh mountall.sh mountall-bootclean.sh
-urandom: hwclock.sh mountall.sh mountall-bootclean.sh
-brightness: mountall.sh mountall-bootclean.sh
 alsa-utils: mountall.sh mountall-bootclean.sh mountnfs.sh mountnfs-bootclean.sh
-checkfs.sh: checkroot.sh cryptdisks lvm2
-checkroot-bootclean.sh: checkroot.sh
-screen-cleanup: mountnfs.sh mountnfs-bootclean.sh
-stop-bootlogd-single: mountall.sh mountall-bootclean.sh
-bootmisc.sh: udev checkroot-bootclean.sh mountnfs-bootclean.sh mountall-bootclean.sh mountnfs.sh mountall.sh

I am unclear as to the significance of the reordering of  .depends.* that
happens on the first run. Jesse, is that expected. Does it point to something?

As Ian said previously, we are clearly still missing something here. I am pretty much in
the dark and clutching at straws. But what filesystem are you using? My
/var/cache/pbuilder is ext3.

Sorry not to have a more helpful response at this stage.

Best wishes

Mark