On 1/3/23 10:27, Jim Fehlig wrote:
On 1/2/23 08:26, Andrea Bolognani wrote:
On Fri, Dec 23, 2022 at 11:56:03AM -0700, Jim Fehlig wrote:
%package daemon-qemu
Summary: Server side daemon & driver required to run QEMU guests
+ %if %{with_modular_daemons}
+Requires: libvirt-daemon-common = %{version}-%{release}
+Requires: libvirt-daemon-log = %{version}-%{release}
+Requires: libvirt-daemon-lock = %{version}-%{release}
+Requires: libvirt-daemon-plugin-lockd = %{version}-%{release}
+ %else
Requires: libvirt-daemon = %{version}-%{release}
+ %endif
This version is fine, but as explained elsewhere I think it would be
better to have
Requires: libvirt-daemon-common = %{version}-%{release}
Requires: libvirt-daemon-log = %{version}-%{release}
Recommends: libvirt-daemon-proxy = %{version}-%{release}
and no dependency at all on the locking part.
Rationale:
* virtproxyd being present allows clients that are older than ~2
years to connect, so it should be there by default while still
making it possible for the admin to opt out, which can be done by
simply uninstalling the corresponding package;
Agree. virtproxyd was created to help transition from monolithic to modular
daemons so makes sense to include the weak dependency.
* storage locking is not the default behavior and needs to be
turned on explicitly, so it's not a big deal if part of the setup
involves installing a couple extra packages in addition to
editing some configuration files, and everyone else gets a leaner
installation.
I'm fine dropping the daemon-lock dependency. I do seem to recall an old
discussion about enabling lockd by default, but I guess it's no longer necessary
with qemu locking image files these days.
BTW, I've made these changes to a local branch and pushed to my fork
https://gitlab.com/jfehlig/libvirt/-/tree/spec-modular-daemons-v2
and enabled the CI pipeline when pushing
https://gitlab.com/jfehlig/libvirt/-/pipelines/737202454
Let me know if a V6 is needed. This patch is the last one without a R-B.
Regards,
Jim
