On Thu, 11 Jun 2015 15:56:20 Andreas Beckmann wrote: > similar problems in the gfmd package.
I'm still sceptical about this. :(
> On 2015-06-10 01:10, Dmitry Smirnov wrote:
> >> 0m43.5s ERROR: FAIL: Package purging left files on system:
> >> /var/lib/systemd/deb-systemd-helper-masked/ not owned
> >> /var/lib/systemd/deb-systemd-helper-masked/gfsd.service not owned
> >>
> >> IIRC these symlinks should be managed by init-system-helpers
> >> (and that packages was installed at purge time for this test).
>
> (and that prevents false positives)
>
> > How could that be if package handles its service file using dh_systemd
> > helper? I very much doubt this problem really present in the "gfsd"
> > package. It looks like false positive or maybe a problem in
> > "dh-systemd"...
>
> There should be quite a number of packages using dh_systemd (I would
> hope), but I only see leftover symlinks in very few of them, so maybe
> it's the way dh_systemd is being used in these packages? Or maybe they
> use features not used elsewhere?
I do not use anything special of dh_systemd. Packaging do not mess directly
with folders like `/var/lib/systemd/deb-systemd-helper-masked/` etc.
> $ cat debian/gfsd.postrm
> #!/bin/sh
>
> set -e
>
> case "${1}" in
> purge)
> rm -f /lib/systemd/system/gfsd.service \
> /etc/systemd/system/gfsd.service
> ;;
> esac
>
> #DEBHELPER#
>
>
> That postinst does not exactly look like "let dh_systemd handle it",
This is just a precaution: one manually invoked upstream command (setup
helper) in the package can create systemd service file. The above fragment
makes sure manually created service file is not left behind after purging of
the packages.
> maybe the manual rm causes confusion?
Might be but it seems unlikely... I'll have a look.
> Perhaps you should contact the systemd maintainers for further
> investigation and advice.
Thanks.
--
Best wishes,
Dmitry Smirnov
signature.asc
Description: This is a digitally signed message part.
