Hi Joey, Joey Hess <[email protected]> writes: > This is applied. However, I wonder about this part of it: Thanks for applying! Is there a rough time estimate on when the next version of debhelper will be released?
> This is a lot of boilerplate to add to every init script in Debian. > Is there any reason to add this to packages that have no systemd > tmpfiles? No, there isn’t. In other words: If debhelper can do it, you can skip the boilerplate unless the package ships any files in /etc/tmpfiles.d or /usr/lib/tmpfiles.d. >> +# In case this system is running systemd, we make systemd reload the unit >> files >> +# to pick up changes. >> +if [ -d /sys/fs/cgroup/systemd ] ; then >> + systemctl --system daemon-reload >/dev/null || true >> +fi > > I'm a bit surprised this is necessary. Is this package removal case > currently broken when using systemd with all the packages out there that > don't do this? It’s not exactly broken, but it results in quite a few unnecessary error messages and potential confusion for our users in the case when a service is not stopped on uninstall. mbiebl tested what happens in various cases: http://paste.debian.net/204443/ delete the service file, stop the service, everything fine. http://paste.debian.net/204447/ delete the service file, systemctl status will still show "wrong" state but mention that you should daemon-reload. http://paste.debian.net/204454/ delete the service file, try to issue a restart (fails). http://paste.debian.net/204456/ delete the service file, try to issue a stop (works). In summary: No, not all cases are broken, but there might be some. mbiebl and I agreed that the safest way is to just keep the daemon-reload in there. -- Best regards, Michael -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
