[Top] [All Lists]

Re: slow email validation problems (was reject vs bounce)

2005-09-14 05:42:33

--On onsdag, september 14, 2005 10:26:42 +0100 Paul Smith <paul(_at_)pscs(_dot_)co(_dot_)uk> wrote:

At 22:49 13/09/2005, Harald Tveit Alvestrand wrote:
I'm thinking of generating '250-please wait' multiline response lines
every 30 seconds or so whilst the message is being processed, with a
final '250 OK' or '550 message rejected because of ...' line.

hmm.... "scaling is the ultimate problem", says Mike O'Dell.

If you take 30 seconds to accept a message, and have the default Postfix
process limit of 100, you can accept up to 300 messages a minute - or an
average of 5 per second.

Hmm, but what about RFC 2821 section - which says that after
'DATA termination' (which is what we're talking about) - the timeout
SHOULD be 10 minutes...

That's what the RFC says.

If you're saying that Postfix can't handle that, so that it NEEDS to have
the timeout there set to 30 seconds or less, then all I can say is Hmmm...

I'm saying that if the average time for a transaction (which is under the control of the receiving MTA) increases, there will be a definite impact - which will be most keenly felt by high-volume sites.

More Mike O'Dell quotes, this from the "IETF Guidelines for Conduct", RFC 3184:

  3. IETF participants think globally, devising solutions that meet the
     needs of diverse technical and operational environments.

     The goal of the IETF is to maintain and enhance a working, viable,
     scalable, global Internet, and the problems we encounter are
     genuinely very difficult.  We understand that "scaling is the
     ultimate problem" and that many ideas quite workable in the small
     fail this crucial test.  IETF participants use their best
     engineering judgment to find the best solution for the whole
     Internet, not just the best solution for any particular network,
     technology, vendor, or user.

It might be that on balance, this should be allowed. But consider all the people who will feel the cost of it.