[Raw Msg Headers][Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: The mailq service

> > > > in use by zmailer. I'm running Zmailer 2.99.50-s5.
> > > 
> > > Yes, but you didn't configure your ZMailer with  --with-tcp-wrappers
> > I did. I used the Red Hat spec file for zmailer which has it by default.
>   Sorry, we were interleaving two question/reply exchanges, thus
>   you (and I) became to repeat question, and answers..
OK. The usual stuff ;_)

> > > How about -lwrap ?
> > Linked in.
> Then I can think of only a tcp-wrapper usage snafu:
> Perhaps you have an  "ALL: ALL@ALL" entry in the  hosts.allow  in
No. Both hosts.allow i hosts.deny contain only specific entries for
specific services and/or hosts...

> The thing is, tcp-wrapper does not work in 'first match only' basis,
> rather it scans thru the file to do 'any match' (to allow/deny)
> Or at least the way the ZMailer uses it is 'any match'.
I think it doesn't apply here, meaning that there are no ambiguities in
these config files... Perhaps I should recompile zmailer with the latest
Alan Cox' update to tcp_wrappers?

> > Excellent! Speaking of features. Is there anything in the plans about
> > adding any capability for the pop3/imap servers to use zmailer aliases (I
> > mean other than modifying them to read the databases directly)?
> Depends on what you want to do with such co-working ?
> What would be the benefit of having the aliases accessible to the
> pop3/imap servers ?
I run several virtual domains on my host and their users have to configure
their MUAs to receive mail from the real account they have on my host (the
virtualization isn't full - to save disk space), so instead of


they have to use


in their POP3 clients. This can get a bit tiresome with time and doesn't
look very well. The full virtualization of the mail system would allow
them to use virtual_user@domain.com to receive their mail via POP3
connections from the real host machine. Of course, it's a matter of
convenience to the user, but always a nice one. And having such full
virtualization takes off some problems from the v-domain administrator. 

> Although, my initial reaction on this was: "layering violation"
> (Message-store access protocols have nothing to do with the message
>  transport systems.)
But they would continue to use the real usernames, those on the real host.
The aliases would be used only at the start of the connection to look the
real user up. At least this is how I imagine it.