Loading ...
Sorry, an error occurred while loading the content.

Re: source & destination based routing: When to decide where to go you need to go

Expand Messages
  • mouss
    ... It would be too complicated to implement this. he would need a first smtpd to do the transport decision, and another one to do the FILTER decision (in a
    Message 1 of 12 , Oct 4, 2006
    • 0 Attachment
      Sandy Drobic wrote:
      > Fabio Pietrosanti wrote:
      >> mouss wrote:
      >>> FILTER applies to a message, not to a recipient, and is overriden by
      >>> FILTER statements in subsequent checks. OP needs a content filter or
      >>> an MDA.
      >> You mean that policy server cannot return to postfix the "FILTER"
      >> statement in order to modify the routing behaviour?
      >
      > No, mouss correctly stated that the filter decision will be decided
      > for the message. In case of multi-recipient messages the mail for all
      > of these recipients will then be routed according to the filter
      > statement.
      >
      > That means if you want to filter with the filter action you should
      > make sure to define a transport that takes care to separate
      > multi-recipient messages into several single-recipient messages.
      >
      > <transportname>_destination_recipient_limit=1
      >

      It would be too complicated to implement this. he would need a first
      smtpd to do the transport decision, and another one to do the FILTER
      decision (in a single "step", FILTER is performed before transprt so you
      can't mix their effects).

      a content_filter is better.
      Note: He can use maildrop/procmail as a content_filter...

      >
      >> Postfix ask what to do to the policy server by passing:
      >> recipient=x@...
      >> sender=y@...
      >>
      >> Policy server check in my database with a sql statement like 'select
      >> outgoing_gateway from routing_tables where recipient=x@... and
      >> sender=y@...' .
      >>
      >> Policy server return as action FILTER smtp:host:port

      smtp:[ip]:port is better.

      >>
      >> Does this could work?
      >
      > Yes, that should work, provided you only have single-recipient
      > messages at that stage.
      >
      > Sandy
    Your message has been successfully submitted and would be delivered to recipients shortly.