[etherlab-users] Handle slave reboots once the master has been activated

[Gavin Lambert]

On 21 March 2018 21:21, quoth Joye Laurent:
> Is there any chance to handle slave reboots during the realtime operation
> (ecrt_master_activate has already been called)?
> I've seen that, once a slave reboots, the master indicates a working counter
> change and the slave is automatically reset in OP mode.
> 
> Is it possible, from my master user space application, to continue to
> communicate with the freshly rebooted slave device, without deactivating
> the master?

Yes, that happens automatically -- it's why the slave returns to OP after reboot, because it gets reconfigured and then starts working normally again.  The working counter should return to normal after this.

You do have to be careful to set up your slave configurations the correct way, however -- in particular you can only use pure position-based addressing if you have a simple chain network, such that when a slave is unplugged or rebooted everything with a higher position also disappears from the network until it returns.  If you have any tree branches in your network, such that some slaves will remain online but change positions when one slave reboots, then you need to use alias-based addressing instead (or a hybrid, where you use an alias address for the first slave in each "branch" and positional addressing for each slave after that in the same branch).