indicator-session-service crashed: Unable to get system bus

StacktraceTop:
 users_service_dbus_init (self=0xd78820) at /build/buildd/indicator-session-0.2.93/./src/users-service-dbus.c:153
 g_type_create_instance (type=<value optimized out>) at /build/buildd/glib2.0-2.29.10/./gobject/gtype.c:1885
 g_object_constructor (type=<value optimized out>, n_construct_properties=0, construct_params=0x0) at /build/buildd/glib2.0-2.29.10/./gobject/gobject.c:1629
 g_object_newv (object_type=140346109475056, n_parameters=0, parameters=0x0) at /build/buildd/glib2.0-2.29.10/./gobject/gobject.c:1412
 g_object_new (object_type=140346109475056, first_property_name=0x0) at /build/buildd/glib2.0-2.29.10/./gobject/gobject.c:1322

Some of the duplicate bugs were filed on Precise, and had the same backtrace. So I'm marking this Confirmed since the problem still seemingly exists with the latest versions of indicator-session.

This used to occur when an instance of indicator-session would try to register itself on the system bus but would fail for some reason -- say, if there was already an instance running that hadn't let go yet, or if there was an issue with the bus. Variations of this crash message happened for various indicators.

This was fixed in 13.10 by demoting these messages from g_error() to g_warning() so that a crash wouldn't occur, triggering a launchpad report and exiting gracefully. It's then the responsibility of the manager that launched the indicator to wait and try again. This is currently being done by ted's upstart work on the indicators.