On 14/03/2013 18:28, /dev/rob0 wrote:
On Thu, Mar 14, 2013 at 11:47:18AM +0000, Elaconta.com Webmaster wrote:
Elaconta.com Webmaster wrote:
Benny Pedersen wrote:
Elaconta.com Webmaster skrev den 2013-03-14 11:48:

Specifically, if a user sends 100 emails and more than 25 of those
are send to non-existing users, disable email relaying for that user
for half an hour, for instance.
i say reject_unverified_recipient one more time

should i give links to pypolicyd-spf ?

sending 100000000 emails does change spf records ?

if you want another way of solving use rsyslog with sql query
>from tempfails/hardfails on random recipient domains, and from
that sql logs block that sender local that spammed to it,
impossible to use reject_unverified_recipient or just ignoreing
the problem ?
I'll look into reject_unverified_recipient, thanks for your
insight.
That's abusive toward other sites. I wouldn't recommend that.

Also for more clarification: We require authentication for all of
our email users, and have hourly email sending quotas in place.
I don't think your quotas are strict enough, and rather than being
renewed hourly, perhaps it should be a one-time trigger to block all
further sending by those credentials. Bear in mind that these aren't
your users sending mail to their friends -- this is malware which has
seized your users' machine and is using their stored credentials to
spew spam to victims. Yes, you will inconvenience your users, but
surely their malware has violated their ToS? (Rewrite your ToS if
necessary. Hold users accountable for such abuse.)

But there's nothing stopping auth'ed users from sending emails to
lots of non-existent users, and that affects the email server's
reputation negatively.
Once abuse has occurred, you can bet more will follow. Don't allow
spew to continue.

Hence our trying to reduce to amount of emails send to non-existent
emails.
I think you'd also find that URIBL content filtering of submitted
mail very effective. This malware almost always tries to get victims
to look at spammy links.
Yes, skimming through the reject_unverified_recipient, i got the gist that SAV is kind of a bad thing. Not going to use it. I'm going to try and implement some URIBL content filtering for outgoing emails, thanks.

Reply via email to