How to set up ring topology?

EtherCAT in industrial settings is often configured in a ring topology, where the master has two EtherCAT ports (In and Out), and its Out port then connects to the In port of Slave 1, whose Out port then connects to the In port of Slave 2 etc until finally the Out port of SlaveN connects to the In port of the master and the loop is closed. This provides a degree of redundancy against dropped connections and also makes synchronization of the clocks slightly neater because all the messages flow in one direction around the ring instead of bouncing back from the final unterminated slave all the way back the whole linear chain (I do realize that in practice the linear topology with the messages bouncing back also works very well).

If my linuxCNC PC has several Ethernet ports, can I define two of them as EtherCAT In and Out ports to achieve a ring topology, and if so, what do I need to define in the ethercat_conf.xml file (and anywhere else) to get this to work? Or is it not supported/too hard?

A basic ethercat_conf.xml begins with something like :
<masters>
  <master idx="0" appTimePeriod="1000000" refClockSyncCycles="1">
... etc.
How does this need to change to achieve a ring topology? And indeed, where does one tell it which ethernet port is performing which function (EtherCAT In, or Out etc.)

Many thanks!

James

There's a config for redundant Ethernet interfaces in /etc/ethercat.conf, but it's not at all clear how that's supposed to be wired up. I suspect if you connect the "backup" interface to the end of your EtherCAT chain that it'll end up *acting* like a ring, but I wouldn't be surprised if something slightly odd ends up happening. The Etherlab docs don't spend much time on it, unfortunately.

Many thanks. I suspected it wasn’t formally implemented so it sounds like I should stick to the tried and true linear chain topology if I want to avoid exploring too many exotic rabbit holes at the moment…