> Doesn't  the  new  proposed  RFC  allow up to 1024 characters?

ESMTP  commands  may have an ever-larger number of bytes as extensions
are  added.  For  this  reason, ADPHA does not apply its logic to EHLO
sessions, only standard SMTP sessions.

> ADPHA IMHO doesn't not prohibit all E/SMTP hack attacks but rather a
> known  exploit  of  weaker  email  servers.  Wasn't this to prohibit
> client software/hackers injecting multiple DATA command's prior to a
> 220 response?

It  was originally to prevent a number of buffer overflow attacks that
are rarely tried these days (yet anyone who's trying them is certainly
a  black  hat).  For  more  latter-day  situations, it would prevent a
spammer trying to pipeline stuff to your server without a HELO or EHLO
or  some other out-of-sequence sessions, rare but certainly suspicious
occasions.

It's  crucial to note that the ONLY difference between having ADPHA on
and off is that with it on, there will be an automatic blacklist of an
offending  IP;  with  it  off,  any  offending  command  will STILL be
rejected by the server with a 552: Command Exceeds Maximum Length, but
the session will stay up. So turning it off won't fix a broken session
completely,  but  if  a mailer for some reason exceeds the limit, then
backs off and retries after the 552 with a shorter command length, the
session  will be successful. You would see a record of the 552 in your
logs either way, and I'm hard pressed to think of a situation in which
sending overlength commands, then backing off, is normal.

Len's   IMGate  community  has  found  that  when  PostFix  and  IMail
communicate  at  LAN speeds, IMail can detect that overlength commands
have  been  sent  and  blacklist  its  own  gateway,  obviously  a bad
situation.  The  question  of  whether PostFix has a timing issue (not
waiting long enough for a response on a socket, waiting too long for a
response,    things    like    that)    when    dealing    with    the
not-the-fastest-card-in-the-deck  IMail  SMTPD,  or  whether  IMail is
seeing  phantom overlength commands where there are none, has not been
resolved. Len takes the position that PostFix cannot be broken, though
he was not aware of what IMail's function actually does until a couple
of days ago. In the absence of real evidence, I take the position that
the cause of the PostFix alarm is unknown and bears further study, and
certainly  does  not  extend  to  a  blanket condemnation of the IMail
feature.

> Also, Is it good or bad to check/uncheck the ADPHA box and why?

I  don't  think  there's  anything wrong with checking the box if your
IMail  server  is  published on the Internet and is not in a situation
that is known to set off the alarm for currently unknown reasons (that
is, on the same segment as PostFix). It's a bit of host-based security
that,  as  I  mention,  is already in place anyway (only the proactive
blacklisting can be turned on or off) and has been in IMail for a long
time, a default that is active in tens of thousands of installations.

-Sandy


To Unsubscribe: http://www.ipswitch.com/support/mailing-lists.html
List Archive: http://www.mail-archive.com/imail_forum%40list.ipswitch.com/
Knowledge Base/FAQ: http://www.ipswitch.com/support/IMail/

Reply via email to