Tim Litwiller <[EMAIL PROTECTED]> said:

> Darrell, after your fix could you try a setting a machine to send 
> undeliverables to the admin and see if that works.

Hi Tim.  This is the exact beast we are trying to prevent from running
wild.

Here is what currently happens:

'60AllowLocalDomains' is set to allow any e-mail through the system as long
as the domain name matches any of your local domains.  This unfortunately
accepts mail to unknown/invalid e-mail addresses and forwards these messages
through your system for processing.  A very bad practice as for every
unknown/invalid address this mean the message will process through your
virus scanner, qmail, admin mailbox, procmail, spam filter etcetera
generating cpu usage as qmail issues an error report, sends the bounce
message and then the 'the bounce bounced' messages just start to pile in. 
Multiply this by one report where the admin was seeing 5-10 messages per
minute and you see what a nightmare this becomes.  This admin returned after
holidays to > 8000 'the bounce bounced' messages.  Then consider that this
admin pays for Internet traffic!

I have rewrote the 60AllowLocalDomains changing the logic, stated in plain
English from,

check for valid '*@domain'  - to - check for valid 'user@domain'

If the message is for a known/valid user@domain, (the one's we care about)
but for whatever reason is undeliverable, (userquota reached, damaged
mailbox) the normal qmail process will take place.

However, if the message is for an unknown/invalid user@domain (the one's we
don't care about) it is dropped at the smtp connection and not sent for
processing through the system.  The sender receives the standard
smtpd_check_rules error reply generated from the existing '99NotoCatchAll'
fragment.  Qmail does not get involved.  Admin mailbox does not explode with
unnecessary error and the bounce bounced reports.

In addition, as an admin, you may always review your maillog (at your
convenience) to view the unknown/invalid user@domain message traffic being
dropped.

Hope this helps to more clearly explain.

If I may request, I would ask everyone please take a look at my contrib rpm
and become aquainted with your smtpd_check_rules.  My rpm installs only one
file (60AllowLocalDomains) into templates-custom.  Issuing a signal-event
email-update is all that is required to expand the template and test this
new fragment.  Very safe to test.  Very easy to remove and return to the
original fragment.

I have this 'live' on my production server and a few clients that had an
immediate need for a fix.  We are all reporting positive results.

However if _anyone_ has a better idea, I'm always ready to listen and
learn.

Regards,

-- 
Darrell May
DMC Netsourced.com
http://netsourced.com
http://myEZserver.com


--
Please report bugs to [EMAIL PROTECTED]
Please mail [EMAIL PROTECTED] (only) to discuss security issues
Support for registered customers and partners to [EMAIL PROTECTED]
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
Archives by mail and http://www.mail-archive.com/devinfo%40lists.e-smith.org

Reply via email to