On Tue, Apr 03, 2012 at 10:06:05PM -0600, Eric Blake wrote:
On 04/03/2012 06:48 AM, Daniel P. Berrange wrote:
> From: "Daniel P. Berrange" <berrange(a)redhat.com>
>
> It is not safe to restart libvirtd after install of the
> libvirt-daemon RPM, since the corresponding driver modules
> may not yet have been installed by RPM. The solution is to
> move the restart logic to a %posttrans script. This runs
> at completion of the RPM transaction, once everything is
> uninstalled.
> ---
> libvirt.spec.in | 13 +++++++++----
> 1 files changed, 9 insertions(+), 4 deletions(-)
I haven't used %posttrans before, but it looks simple enough. What I'm
not sure of...
>
> diff --git a/libvirt.spec.in b/libvirt.spec.in
> index d867b09..73e1ec3 100644
> --- a/libvirt.spec.in
> +++ b/libvirt.spec.in
> @@ -1188,9 +1188,6 @@ fi
> %endif
>
> /sbin/chkconfig --add libvirtd
> -if [ "$1" -ge "1" ]; then
> - /sbin/service libvirtd condrestart > /dev/null 2>&1
> -fi
> %endif
>
> %preun daemon
> @@ -1207,13 +1204,21 @@ if [ $1 = 0 ]; then
> fi
> %endif
>
> -%postun daemon
> %if %{with_systemd}
> +%postun daemon
> /bin/systemctl daemon-reload >/dev/null 2>&1 || :
> +%endif
is why daemon-reload can be done right away,
This tells systemd to re-read the .unit and .service files, but not take
any action.
> +%posttrans daemon
> +%if %{with_systemd}
> if [ $1 -ge 1 ] ; then
> # Package upgrade, not uninstall
> /bin/systemctl try-restart libvirtd.service >/dev/null 2>&1 || :
but a full restart has to wait for %posttrans. I was naively assuming
they should both be %posttrans.
And this actually makes systemd take some actions.
I guess we could also postpone the daemon-reload to %posttrans without
any ill-effects though.
Daniel
--
|:
http://berrange.com -o-
http://www.flickr.com/photos/dberrange/ :|
|:
http://libvirt.org -o-
http://virt-manager.org :|
|:
http://autobuild.org -o-
http://search.cpan.org/~danberr/ :|
|:
http://entangle-photo.org -o-
http://live.gnome.org/gtk-vnc :|