[etherlab-users] Ethernet and EtherCAT with same type of NIC

[Erwin Burgstaller]

Hi again,

Another thing, so I'm starting an separate thread for it.

The PC we're using as master, has two on-board Ethernet NICs which both
are running with the e1000 or ec_e1000 driver module.

One NIC is for Ethernet, the other for EtherCAT. If I'm working via
ssh on the network and I do /etc/init.d/ethercat start or stop, the
Ethernet goes down in any case, as there's a change of the driver module
and you're locked out. As the maintenance is done mostly via network
connections, this is quite a problem.

There are 2 solutions I've tried.

A) The script /etc/init.d/ethercat has to call 

        /etc/init.d/networking restart 

   after start or stop is done. With good luck the connection is
   reusable after a while. In bad case you have to make a new
   connection.

B) The originally used module must not be used, it's always running
   ec_e1000 in our case and as loading of the ec_e1000 forces ec_master
   to load. So the /etc/init.d/ethercat script isn't needed,
   normally. Unless for stop and then start again of course.

   On Debian systems for example, e.g. the following line:

       options ec_master main_devices=00:D0:C9:AE:4B:41

   has to be added to /etc/modprobe.d/options

   and ec_e1000 to /etc/modules, or maybe better as alias for e1000, if
   that works.

   So as long as there EtherCAT isn't stopped, everything is fine, but
   to be sure, there should the restart of networking in the EtherCAT
   script anyway, but by using ec_* exclusively there's at least no
   broken Ethernet while boot process.


Anyway, I think both this methods aren't very cool, so how could this
problem solved in a better way?

Thanks,
	Erwin

-- 
Erwin Burgstaller