Autor: Jim Cheetham Data: Para: exim-users Asunto: Re: [exim] "unit tests" for an exim installation/configuration
Quoting Sandro Tosi (from 07/12/10 05:07): > So, how are you testing your configuration when developing that "what
> was fixed before is not broken due to new changes" and "what's required
> to works is working"? how do you ensure your configuration/installation
> works when you deploy it in
Most of my tests are based around variations of swaks and netcat -- I
have a set of variations of To:, From:, TLS, Authed, not Authed,
carrying spam (GTUBE), carrying malware (EICAR), too big, from
blacklisted IP, from whitelisted IP, and so on (use 'nc -s' to set the
source address for a connection).
This doesn't check everything, but returns either 'accepted' or
'rejected' early. So one command-line invocation kicks off a couple of
dozen checks.
Deeper tests to make sure messages are actually delivered are just shell
scripts that use swaks to send an acceptable message (with a UUID in the
subject, for example), wait a few moments, then dive into the relevant
maildir looking for that UUID to be on the disk. Then check that it can
be retrieved by IMAPS :-) Some of these are suitable for end-to-end
testing from Nagios or similar.
