Postfix SMTP client issues

damber · *nix forums beginner Joined: 19 Apr 2008 Posts: 2

I've just tried changing to standard HELO/SMTP rather than the default extended command set and it does separate out the commands individually - but the problem still occured Sad

I then decided to change the IP address the smtp out binds.... and yup.. godaddy sucks big time. it only allows 1 of the 3 ips to actually send mail..... and none of their staff seems to know this, as I asked this question (amongst many others) and got the usual "it should work - your server is at fault".

ah well... Smile

damber · *nix forums beginner Joined: 19 Apr 2008 Posts: 2

I'm trying to get postfix setup and working on a dedicated server within godaddy's network, which must relay through their smarthost/smtp relay server.

Unfortunately I keep getting "553 Sorry, that domain isn't in my list of allowed rcpthosts. (in reply to RCPT TO command))"

So, I tested that the server could actually send mail through this relay server using "telnet k2smtpout.secureserver.net 25" (godaddy's smtp gateway), and it worked fine.

I checked that all the DNS records were ok (with correct MX records etc) in case that was being validated etc, I also checked that the same IP was used in the manual connection as the postfix connection.. both ok.

To check what was being 'said' during the event I did a tcpdump of the session and compared the postfix session and the telnet session outputs.

This resulted in my seeing that THE EXACT SAME COMMANDS are issued (from ehlo hostname to MAIL FROM, RCPT TO, DATA etc) from both sessions... so, the same DNS records, the same hostname, the same source/target emails, the same commands etc were all being used.. but one works, and the other doesn't.

The ONLY difference I can see from the packet trace is that the automated session sends the MAIL FROM, RCPT TO and DATA commands in a single packet separated with crlf's (0d 0a), whereas the manual session sends the packets individually (as you would expect as each crlf in the telnet client issues a packet).

I am running CentOS 5 with Postfix v2.3.3. Godaddy's smtp server (I think) runs qmail, and doesn't require SMTP auth (I assume it uses the IP / network of the server, seeing as it's dedicated and hosted by them, they can trust it to some degree.

As this is the only difference between the packets, I can only assume this is the issue... so a) am I right about that, or do you have any other ideas how/why this bizarre behaviour might occur and b) how do I resolve this problem? Can I force postfix to send the commands individually rather than in a single packet? or is there some configuration flag to fix this type of scenario ?

if you want to know a particular part of the postconf output or other trace/config info let me know (not included for brevity).

Thanks for any help you can give.

Google