Auteur: Drav Sloan Date: À: Marc Perkel CC: exim-users, Dean Brooks Sujet: Re: [exim] Retry on 550 errors
Marc Perkel wrote: > One of the problems is that when you're in the spam filtering business
> if a spam gets through then it appears that I'm the source of the spam.
> People like gmail, yahoo, hotmail, comcast are also getting blacklisted
> every now and then.
And then going on to break RFC legitimises this blacklisting. I've worked
for big ISPs, and when the smarthosts got blacklisted, it was my
responsibility to fix that before too much collateral damage was done.
I may have routed things differently (maunal routes) to avoid complaining
MTAs, or froze messages automatically until such times as the issue
was fixed.
> Then there are companies that are so swamped with spam that they change
> the rules and I get caught in the rule change. Godaddy had a server of
> mine blacklisted yesterday because my HELO didn't match my IP.
Then Godaddy are not aware of email standards which (in basic terms)
say that the HELO value is not authorative and therefore is not really
something you can take actions on.
> Then
> there's SPF which totally sucks where the customer's server rejects
> email because SPF doesn't match when I'm forwarding email. So one
> customer is losing their Netflix email.
SPF is already notorious for not being a complete system, in that it
doesnt work for hosts who do forwarding for other domains and so on.
However once again, contact the admin of the host with the SPF in place
and get them to allow exceptions to their policy.
> Email protocols need to be rewritten.
We all know this, but I can't see it happening quickly.
In the interim breaking RFC by ignoring 5xx's is NOT the way to
go; EVEN in a closed environment like Marcs. He has plenty of
other options available; manualroutes, freeze the email, batch
SMTP it, route via another server he owns which may not get the
issue talking to those mail servers.
I strongly dislike 'hacks' which fly in the face of RFC standards;
especially ones put inside exim's distributed source code, because
it proliforates what you argue as a "small case" for use into something
anyone with exim can do; which is a bad thing [tm].