Plugins » Foreman Remote Execution

Users are also going to want to pick a particular interface - not necessarily the primary - for the remote execution proxy to talk to, so that adds another layer of complication.

Are we talking about the mode, when `remote_execution_global_proxy` is truned on or off http://theforeman.org/plugins/foreman_remote_execution/0.0/#4.1Determiningthesmartproxyforhost ?

When the seeting is set to false, it should behave as you described (unless we have a bug in the code), but if I understand it correctly, you've just described what we have written in the manual.

FYI here is the code that should make that happen https://github.com/theforeman/foreman_remote_execution/blob/master/app/lib/actions/remote_execution/run_host_job.rb#L69-L90

I guess I should RTFM, that's partially it.

The logic doesn't make sense to me though, and I don't think meets my requirement here in this ticket.

Subnets have DNS, DHCP and TFTP proxies, which may all be the same, or maybe all different, with varying levels of connectivity to hosts. The code just picks whatever random one of those has a remote execution provider (e.g. SSH).

There should be a proper Remote Execution Proxy association on the subnet. You might run into the problem where you want a SSH and Salt provider in the same subnet. That could be a has_many association, or a tradeoff we have to accept (one provider per subnet).

Yes, I agree on that: the reason for the current behaviour is: it doesn't go against the users at the beginning
(which is IMO important to help users with starting using it), and in many cases, the behaviour will just
work.

Of course, the case you're describing is valid and I'm all for solving that.