home | O'Reilly's CD bookshelfs | FreeBSD | Linux | Cisco | Cisco Exam  


sendmail

sendmailSearch this book
Previous: 10.6 Things to Try Chapter 11 Next: 11.2 Rule Set S=
 

11. Rule Sets 1 and S=

Recall that the client.cf file that we have been developing has two purposes: to forward all mail to a central mail hub for processing and to make all mail appear as though it originated from the hub. The first purpose was fulfilled by the creation of a rule set 0 that always selects the hub delivery agent. In this chapter we begin the process of fulfilling the second purpose, by designing rule sets 1 and S= .

But first, to review, recall also that the recipient's address is first processed by rule set 3 and then by rule set 0. Rule set 3 finds an address among other text and cleans up the address by removing nested angle brackets. After that, rule set 0 selects a delivery agent on the basis of the recipient's address. If the address is bad, rule set 0 selects the error delivery agent. For a valid address, rule set 0 selects the hub delivery agent. Once the delivery agent is selected, the processing of other addresses, such as the sender's address, may proceed.

11.1 Flow of the Sender's Address

When sendmail begins processing a mail message for delivery, it first looks for the envelope [1] recipients. Each recipient address that is found is first processed by rule set 3, and then by rule set 0, which selects a delivery agent.

[1] See Chapter 1, Introduction , for a review of the concepts of header versus envelope.

After appropriate delivery agents have been selected for all recipients, sendmail processes the sender's address. There is usually only a single sender [2] for any given mail message. The sender's address (or addresses) may appear in the envelope or in a From: header line, or it may be derived from the uid of the process that ran sendmail .

[2] The envelope sender and the header sender are not necessarily the same.

As shown in Figure 11.1 the address is processed by rule set 3 first and then by rule set 1. Then, for each recipient in the list of recipients, each sender's address is custom processed by the rule set specified in the recipient delivery agent's S= equate. This custom processing is necessary because different delivery agents may want the sender's address to have a different appearance. One example is the difference between a domain-based delivery agent (needing an address such as gw@wash.dc.gov ) and a UUCP-based delivery agent (needing a routing path such as fbi!wash!gw ).

We will examine the custom S= processing first, then discuss why a generic rule set 1 is not needed in the client.cf file.

Figure 11.1: The flow of the sender's address

Figure 11.1