Comment 3 for bug 1893669

The "SERVICE_GROUP_AGGREGATE" is missing for the service groups of oam-services, controller-services, cloud-services, and vim-services in DX. The change was recent, only in DEV branch.

https://opendev.org/starlingx/stx-puppet/commit/fbb4cdef07c52acb66cfcaf91bc2d029ffb00ff1

Because the service_group_aggregate is missing, when the service failed, instead of rescheduling all aggregated service groups to peer controller (swact), the impacted service group (controller-services) is rescheduled to the peer controller, which also failed (b/c it depends other services, which are scheduled in different node).