>How common is this? Why did postfix use the lowest MX priority server
>for=20
>hawaii.rr.com to try and verify?

Let's assume, and it's has always been safe ime, that postfix follows the 
"MX algorithm" rigorously.  Contacting the last MX would mean that all 
higher preference MX's failed to respond at all (ie, no 4xx or 5xx, but 
with timeout).

hawaii.rr is probably not a huge infrastructure and maybe very stressed, 
ie, slow.

if this is large problem, then you could try increasing the smtp timeouts:

smtp_connect_timeout = 30s
smtp_data_done_timeout = 30s
smtp_data_init_timeout = 30s
smtp_data_xfer_timeout = 30s
smtp_helo_timeout = 30s
smtp_mail_timeout = 30s
smtp_quit_timeout = 30s
smtp_rcpt_timeout = 30s
smtp_rset_timeout = 120s

context: you can see in pflogsumm report that nearly all mail passes 
through IMGate in under 10 secs, so having "smtp_connect_timeout = 30s" is 
not too BOFH, imo.

>   Would this be why verification failed?

apparently, the only MX that responded refused connection on port 25.  If 
you were postfix, what would you do, after you had already timed out trying 
to even get a refusal at the others?

man verify

... is worth reading.

>Also, is there a way to remove this address from the
>address_verify.map.db? (or add it to the db, if it's used as a
>whitelist)

man postmap

shows no "delete" function.

There must be some tool for this, eg, I know Bennett Todd's pop-before-smtp 
PERL script deletes records from its .db file when an ip's POP3 login grace 
period expires.  but I don't know one.

Len


Reply via email to