| |||||||||||||||||||||||||||||||
|
> Wietse Venema: > >>Jacqui caren: >> >>>I have been processing and forwarding email using a perl >>>script and SMTP however this eats resources. >> >>Don't execute one script per message. Use a persistent process. >> >> >>>What I would like to do is once I have processed the message >>>I would like to use /usr/lib/sendmail -g -I or similar >>>to deliver into the maildrop queue. >>> >>>However SMTP gives me a message ID that I use for an audit >>>trail. The /usr/lib/sendmail delivery mechanism >>>does not seem to have that option. >>> >>>FWUI /usr/lib/sendmail is a front end for an undocumented >>>internal protocol (postdrop?). Could I write a perl module that >>>would deliver the message using this protocol (and give me >>>a message id etc) or is this protocol "subject to change >>>without notice"? >> >>Using internal interfaces is unsupported. They change without >>notice, and will break with the next release. > > > It is possible to add a published submission interface to postdrop, > but it still would not give control over the queue file name that > the daemonized portion of Postfix will use. > > You can, however, submit mail into /usr/sbin/sendmail with your > own unique Message-ID header, and use that for tracking. Reasonable > mail systems, even those that don't implement the DSN protocol, will > report the Message-ID information in an NDR. It is more a case of clients complaining that they have received thier urgent email sent less than five minutes previously. We have to (from the database logs) track through to the message queue id and then use this to find out when it was submitted (known in DB) and delivered or bounced etc. For SLA we need to be able to *prove* an attempt to deliver was made within one hour of generation and we did not withhold delivery.
| ||||||||||||||||||||||||||||||
© 2004-2008 readlist.com