A few misc commits to make things nicer:
-T (timeout) now works for ansible-playbooks, just like it did for /usr/bin/ansible. If you have hosts that are offline -T 5 is much less waiting than the default. I’m thinking of lowering the default down to about 10 seconds anyway. Seems reasonable, right?
Failed hosts in playbooks are not tried again later on in the playbook run, even if that failure is in setup or async mode. This means you only wait for that SSH timeout once, not more than once.
Global error handler in the core such that any traceback in running one node doesn’t stop other nodes, general error handling cleanup. This means the “dark hosts” message is about more than SSH timeouts, it gathers up any exceptions you may have had that are not clean returns from the module.
Tree mode fixed. --tree mode now works again to save results files in a file per host, meaning it no longer saves everything in multiple files.
–Michael