Bug#1039361: SAUCE systemd unit
Hello,
On Mon 08 Dec 2025 at 01:44pm GMT, Ian Jackson wrote:
> Ian Jackson writes ("Re: Bug#1039361: SAUCE systemd unit"):
>> Sean Whitton writes ("Bug#1039361: SAUCE systemd unit"):
>> > It looks like the problem is that the 'exit 0' in /etc/sauce/sys-config,
>> > that means the service doesn't start.
>>
>> That exit 0 is deliberate, since it shouldn't run until it's been
>> configured.
>
> Also:
>
> It works with the init script, even under systemd.
>
> The test case includes rebooting the simulated host, and then checking
> that the sauce daemon seems to be running by the time the harness
> resumes the test post reboot. t-startup, lines 60-64.
Ah. You need to do 'systemctl enable sauce' for it to come up after a
reboot. Maybe that's the problem.
--
Sean Whitton
signature.asc
Description: PGP signature
Bug#1039361: SAUCE systemd unit
Ian Jackson writes ("Re: Bug#1039361: SAUCE systemd unit"):
> Sean Whitton writes ("Bug#1039361: SAUCE systemd unit"):
> > It looks like the problem is that the 'exit 0' in /etc/sauce/sys-config,
> > that means the service doesn't start.
>
> That exit 0 is deliberate, since it shouldn't run until it's been
> configured.
Also:
It works with the init script, even under systemd.
The test case includes rebooting the simulated host, and then checking
that the sauce daemon seems to be running by the time the harness
resumes the test post reboot. t-startup, lines 60-64.
Ian.
--
Ian JacksonThese opinions are my own.
Pronouns: they/he. If I emailed you from @fyvzl.net or @evade.org.uk,
that is a private address which bypasses my fierce spamfilter.
Bug#1039361: SAUCE systemd unit
Sean Whitton writes ("Bug#1039361: SAUCE systemd unit"):
> The use of Type= and RemainAfterExit= is correct, I think.
>
> It looks like the problem is that the 'exit 0' in /etc/sauce/sys-config,
> that means the service doesn't start.
That exit 0 is deliberate, since it shouldn't run until it's been
configured.
It's edited away by the autopkgtest. debian/tests/t-startup, line 42.
Ian.
--
Ian JacksonThese opinions are my own.
Pronouns: they/he. If I emailed you from @fyvzl.net or @evade.org.uk,
that is a private address which bypasses my fierce spamfilter.
Bug#1039361: SAUCE systemd unit
Hello,
On Sun 07 Dec 2025 at 10:59pm GMT, Ian Jackson wrote:
> Ian Jackson writes ("Bug#1039361: SAUCE systemd unit"):
>> OK, I guess I will have a go at this myself this weekend.
>
> I converted the package to use dh_installinit and dh_installdeb.
> That's in 0.9.5 which I just uploaded.
>
> I wrote a service file based on the systemd documentation I could
> find, which calls the existing init script. (I think this is probably
> the best approach, but I'm not sure.)
>
> But it does not work. My autopkgtests fail. I'm running
>
> git-clean -xdff && dpkg-buildpackage -uc -b && autopkgtest
> ../sauce_0.9.5~~_all.deb . --- /usr/sbin/really autopkgtest-virt-qemu
> /volatile/ian/autopkgtest-virt-qemu/forky
>
> where that image came from really autopkgtest-build-qemu forky forky.
>
> It's evidently running my unit file, since that's the change to
> the .deb and without it everything works, presumably via systemd's
> init script compatibility report, whose planned removal is what this
> bug is about.
>
> I'm not sure how to debug this. I'm not a habitual user of qemu, so
> I' m not sure even how to log into the qemu VM, and if I did I
> wouldn't know how to debug a system service startup.
>
> My branch is at
>git clone
> git+ssh://login.chiark.greenend.org.uk/home/ian/things/Sauce/sauce -b
> systemd-draft
>
> My feeling is that converting this package to use the dh sequencer, or
> generally overhauling the rules file, isn't a particularly useful
> thing to be doing. But I could be wrong, especially if that's likely
> to be implicated in the problems I'm having with systemd.
The use of Type= and RemainAfterExit= is correct, I think.
It looks like the problem is that the 'exit 0' in /etc/sauce/sys-config,
that means the service doesn't start.
--
Sean Whitton
signature.asc
Description: PGP signature
Bug#1039361: SAUCE systemd unit
Ian Jackson writes ("Bug#1039361: SAUCE systemd unit"):
> OK, I guess I will have a go at this myself this weekend.
I converted the package to use dh_installinit and dh_installdeb.
That's in 0.9.5 which I just uploaded.
I wrote a service file based on the systemd documentation I could
find, which calls the existing init script. (I think this is probably
the best approach, but I'm not sure.)
But it does not work. My autopkgtests fail. I'm running
git-clean -xdff && dpkg-buildpackage -uc -b && autopkgtest
../sauce_0.9.5~~_all.deb . --- /usr/sbin/really autopkgtest-virt-qemu
/volatile/ian/autopkgtest-virt-qemu/forky
where that image came from really autopkgtest-build-qemu forky forky.
It's evidently running my unit file, since that's the change to
the .deb and without it everything works, presumably via systemd's
init script compatibility report, whose planned removal is what this
bug is about.
I'm not sure how to debug this. I'm not a habitual user of qemu, so
I' m not sure even how to log into the qemu VM, and if I did I
wouldn't know how to debug a system service startup.
My branch is at
git clone git+ssh://login.chiark.greenend.org.uk/home/ian/things/Sauce/sauce
-b systemd-draft
My feeling is that converting this package to use the dh sequencer, or
generally overhauling the rules file, isn't a particularly useful
thing to be doing. But I could be wrong, especially if that's likely
to be implicated in the problems I'm having with systemd.
Ian.
--
Ian JacksonThese opinions are my own.
Pronouns: they/he. If I emailed you from @fyvzl.net or @evade.org.uk,
that is a private address which bypasses my fierce spamfilter.
Bug#1039361: SAUCE systemd unit
OK, I guess I will have a go at this myself this weekend. Ian. -- Ian JacksonThese opinions are my own. Pronouns: they/he. If I emailed you from @fyvzl.net or @evade.org.uk, that is a private address which bypasses my fierce spamfilter.
