[SOLVED] Testing mail server with Telnet - My experience

[void]

*domains changed to protect the guilty*

Howdy folks. This isn't actually a problem, and I do see that somebody else posted it here once before, quite some time ago, but the response was totally incorrect so I am bringing it up again.

Mostly, it's more of an interesting potential bug... Enough preface, here is the issue:


Have you ever tried to test your mail server by using "telnet mail.whatever.com 25" from any kind of command line (windows, linux, whatever)?

You probably have. You probably know already the easy steps to testing that:

helo $domainname
mail from:somebody@somewhere.com
rcpt to:localuser@localdomain.com
data
Blah
.

Standard right?

I was having some issues with email yesterday, and a customer brought to my attention that he took the liberty of testing it via telnet to my server, but that once connected he could do thing.

Here is what happens from Windows telnet:
C:\> telnet mail.mydomain.com 25
220 *****************************
helo asdf.com
402 4.5.2 Error: command not recognized

400 4.5.2 Error: bad syntax
helo sdf.com
402 4.5.2 Error: command not recognized
quit
402 4.5.2 Error: command not recognized
quit
402 4.5.2 Error: command not recognized
quit
402 4.5.2 Error: command not recognized


So, kind of strange. Your first impulse may be like mine was and to say "Your firewall/proxy/what-not device is probably filtering the helo command because only your mail server should be sending mail."

But then why would "quit" also not work? Strange.

I tried again, but this time from a linux machine I have hosted across town at a colo.

[trixbox1.localdomain ~]# telnet mail.mydomain.com 25
Trying 216.1.28.4...
Connected to mail.mydomain.com.
Escape character is '^]'.
220 *****************************
helo blah.test.com
250 mail.mydomain.com
mail fromresident@theworld.com
250 2.1.0 Ok
rcpt to:admin@mydomain.com
250 2.1.5 Ok
data
354 End data with .
Test!
.
250 2.0.0 Ok: queued as 60B11996006B


Strange! This time it worked. In fact, it works from every LINUX machine I can possibly get to. Even from the mail.mydomain.com server to itself. Works fine.

Why won't windows telnet work? I try Penguinet, my favorite little gui SSH program for Windows:

220 *****************************
helo blarh.com
402 4.5.2 Error: command not recognized
ehlo asd.com
402 4.5.2 Error: command not recognized
quit
402 4.5.2 Error: command not recognized

Then I noticed something, if I hit BACKSPACE it doesn't appear to see the correct ^H sequence and draws some gibberish on my screen.

I wonder... is it my terminal?

I connect again but this time I use PUTTY and when it says "Connection Type" I chose RAW:

220 *****************************
helo blarg.com
250 mail.mysite.com
mail from:test@testy.com
250 2.1.0 Ok


It works!

So... something in Postfix hates certain terminals, and this WAS working until I updated Zimbra to version 6. Lets see... ah yes I am indeed on a new version of postfix. Very new. Looks like I hopped from 2.4.7.5 to 2.6.5.2

Cool.

Is email still working though? Yes. Working fine. I get all my incoming mail as far as I can tell. There is nothing in the logs showing "Somebody said HELO in a character set I don't understand or barfed binary gibberish at me during an SMTP session" (ok I made that message up).


So strange.

Maybe a guru out there could shed some light on this little fun mystery. As far as I can tell it's gotta be a config setting in Postfix, cause that is who answers to port 25 with a standard Zimbra install yes?

Anybody else out there notice this odd behaviour too?

[void]

Just for the sake of being insane, I tested this on a few other servers out there running Zimbra.

Same behaviour. Anybody can confirm or deny? All servers are running version 6.XX either Network or Community.

Someone? Anybody test this with their own server and share results?

Feel I should also add the ESMTP inspection is disabled in my ASA 5510.

[void]

Ah ha!

Damn you Cisco.

Hope this helps: login to your firewall and turn off fixup, but you have to do it manually from command line, their GUI doesn't use this feature.
NO FIXUP PROTOCOL SMTP 25

I am no longer greeted with 220 *************** when I login, but instead get the right header, and things work fine.

I hope this small tidbit of knowledge can help somebody else out there, and save them all the googling I just did.