The underlying issue was a switch config problem causing the Ceph storage network to be unresponsive, so no disk I/O was possible for the - still running - U7 VMs. The problem has been solved for now and according to our service monitoring, all VMs with all of their services are running fine again.
