Simple example. I have a host with one interface and I want to reconfigure that interface. Now it can happen that I make a mistake, for example by specifying a IP address without subnet mask (has happened to me a couple of times), then I run the role and lost network connectivity to the host. With the reintroduced ifupdown__reconfigure_auto this can be somewhat limited, but one might not catch all mistakes easily thus it would be nice if the script which applies the interface config would roll back to the last known good state if something goes wrong.
The question is just how to detect that. I see two options:
Interface can not be brought up because of an error. Every simple to detect.
No connectivity to the Ansible controller. Maybe try to ping the Ansible Controller which configures the remote host and if there is no response, assume something has gone wrong, of course, allow to influence this detection by inventory variables.
Simple example. I have a host with one interface and I want to reconfigure that interface. Now it can happen that I make a mistake, for example by specifying a IP address without subnet mask (has happened to me a couple of times), then I run the role and lost network connectivity to the host. With the reintroduced
ifupdown__reconfigure_autothis can be somewhat limited, but one might not catch all mistakes easily thus it would be nice if the script which applies the interface config would roll back to the last known good state if something goes wrong.The question is just how to detect that. I see two options:
Related to: https://github.com/debops/ansible-ifupdown/pull/58#issuecomment-260877767 PS: I really liked pyroman because of that feature :wink: