Could you clarify what exactly you mean with configuring a slave? Do you mean setting parameters in a single slave? In Ethercat there are so called "start up parameters" that are automatically written to a slave while boot up. See example in this post.
Like mentioned the EtherCat master in the actual state will give addresses automatically depending on the topology. See online documentation.
I think you will need special EtherCat switches because of the technology used.
Not EtherCat TCP packages are suppressed.You will need to use "mailbox" protocolls (see docu) like EoE (Ethernet over EtherCat)
The hardware topology can be a line or star. Internally in switches it will be reconfigured to a line again. Some topology discussion can be found already in this post.
You can also switch up the EtherCat partially to Pre-OP and not all modules need to attached. In standard all data is transmitted in one package and as soon as one datum is faulty all data will get invalid. But several data packages can be set up that are checked individually. This is handled via so called "sync units" (see docu). So if a missing slave is another sync unit then the rest, the bus can go up.
Beside the scan function you can of course also add build up your IO project "offline" and then only load it once to the EtherCat master.
See for further information the:
online documentation of the ctrlX CORE - EtherCat master app
Collection of How-to videos, blogs and examples for ctrlX AUTOMATION
... View more