Temporary errors are NOT permanent

I’ve had some complaints recently of greylisting generating bounces.  This should not be happening, so until this morning I made this off as total crap stories that people are pulling on me in order to once more receive their daily dose of spam.  As it turns out, however, there are in fact some large govermental organizations that believes that temporary errors are in fact permanent (And probably others too).

I’m rather not going to mention exactly who the proven braindead department is, but let’s summarize it as follows:

  1. user @ department braindead sends email to my client.
  2. my mail server gives 451 Greylisting (INIT) when their relay tries to deliver.
  3. Some random time (up to 10 minutes later) their relay retries again, and since time limit is 15 minutes they get 451 Greylisting (TIME).
  4. Their relay generates a bounce.

Step 4 is a problem.  If they treat all temporary errors this way it would mean that if a server is temporarily down with a real problem (like disk full, or the MTA died, or AV isn’t available, or load is too high that it times out, or any of a large number of other possible problems) their users are going to after 10 minutes get a bounce stating that the message delivery has failed.  And this is exactly what we have observed.

The bottom line is that this is a problem with the sending server incorrectly treating temporary deferrals as permanent errors in my opinion.  When we requested the user to resend the message after the bounce it came through immediately as was expected.  And will now continue to do so permitting she sends at least one message to our client every 30 days.  Not unimaginable.

Comments are closed.