Spam Attacks

[Glenn Steen]

On 14 September 2011 12:07, Paul Kelly :: Blacknight
<paul at blacknight.com> wrote:
> <snip>
>
>> I've got pflogsumm daily reports stored since August -08, and apart
>> from natural differences (layoffs making the total volume drop,
>> temporarily driving up the "no such address"-rejections back in -09) a
>> cursory comparision of a semi-random selection show no real difference
>> during the last months (well, there's always a lull during the
>> summer/vacation period:-).
>>
>
> [Paul Kelly] Care to share percentages maybe? :)

:-)
Well, over the entire period, all rejections seem to lie within the
span 30-60% of all incoming delivery attempts, and for the last couple
of months... closer to 30 than anything.
I don't have the legal ability to use RBLs for rejections, only points
in SA, due to specific Swedish legislation, so all rejections come
from "RFC strictness", relaying attempts, fraud (pretending to be a
local sender) and recipient verification. As Martin notes, the last
bit seem to grab what you're seeing, so I should be seeing the same
increase, but match on different rejection criteria.
Since the layoffs (reduced the workforce by a third), our domain
(ap1.se) is real low volume, and ... luckily has escaped attention so
far:-).

>> But that  might only show quirks of my particular setup, volume and
>> usage patterns of my userbase etc. I suspect an ISP-type organization
>> would be more likely to ... attract ... badness:-).
>
>  [Paul Kelly] Believe it or not, 10 domain names from the 43k or so hosted on our Qmail cluster get 94% of all e-mail. One of them is a medical research company. I suspect someone out there doesn't like them :-)
:-)
>
> Top 10 domains received 20,000,000 (yes 20 million) + spam rejections in the last 10 days.
>
.... Oh ...
> It's a wee bit crazy.
>
Yep, real crazy. How is the PF things holding up under the deluge?

> Regards,
>
> Paul
>

Cheers!
-- 
-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se