On Thu, Sep 29, 2011 at 11:45 AM, Daniel P. Berrange <berrange(a)redhat.com&gt; wrote: > On Wed, Sep 28, 2011 at 09:30:57PM +0200, Marc-André Lureau wrote: >> --- >>  libvirt-gobject/libvirt-gobject-connection.c |  126 ++++++++++++++++++++++++++ >>  1 files changed, 126 insertions(+), 0 deletions(-) >> >> diff --git a/libvirt-gobject/libvirt-gobject-connection.c b/libvirt-gobject/libvirt-gobject-connection.c >> index e99a08d..34781e6 100644 >> --- a/libvirt-gobject/libvirt-gobject-connection.c >> +++ b/libvirt-gobject/libvirt-gobject-connection.c > >> @@ -277,6 +398,10 @@ gboolean gvir_connection_open(GVirConnection *conn, >>          return FALSE; >>      } >> >> +    if (virConnectDomainEventRegister(priv->conn, domain_event_cb, conn, NULL) == -1) { >> +        g_warning("Failed to register domain events, ignoring"); >> +    } >> + > > You might have noticed that we jump through some horrific hoops in > gvir_connection_fetch_domains(). The reason for this is so that we > can issue fake event notifications for start/stopped/added/removed > when virConnectDomainEventRegister() is not available. Not all hypervisors > support the virConnectDomainEventRegister() API, so we need that fallback > code. When you succesfully call virConnectDomainEventRegister(), we should > disable the event emitting part at the end of gvir_connection_fetch_domains() > to avoid duplicate events. In that case, create_domain() method should emit the signal too when event didn't register?