We had a stray letter in a configuration field intended to be used
for port numbers which resulted in port numbers such as 587c where
the letter tripped up Postfix' SMTP client which in turn caused
it to exit early. Any e-mail stuck in queue to be sent upstream would
remain stuck indefinitely as no connection was possible.
We're opting to not implement a config test. A 'postfix check' will
not find such issues and a service restart via an Ansible handler
will work without issues. Postfix will only complain about such an
issue when connecting its client to the upstream server for real mail
delivery. Any detection or validation mechanism would require the
same real-life test of sending an e-mail. At this time its appears
saner to not implement a test rather than implementing an invasive
test.
