In message <9A251497-E94C-4693-8E89-3FD3ACF6D138@stupendous.net>, Nathan Ollere nshaw writes:
On 24/01/2009, at 6:46 AM, Steven Lisson wrote:
Hi,
I agree with seeing no traffic to/from 66.230.128.15 but am still seeing flows 'from' 66.230.160.1
Regards, Steve
Hi Steve,
There is at least an iptables rule you can use to drop this specific query, assuming your nameservers run linux.
http://www.stupendous.net/archives/2009/01/24/dropping-spurious-nsin-recursi... e-queries/
The bind-users mailing list suggested having the ISPs trace back the flows and find the networks emitting the spoofed packets, and have those networks implement BCP 38.
It was also said here.
While that's the 'right' solution (everyone should be doing ingress filtering, sure, impossible to argue against it), not every network out there is operated by people who give a damn.
I would suggest that you don't want to peer with such networks. I would suggest that deploying BCP 38 be a requirement for peering.
This will work at least until the kiddies improve their scripts to query for names that actually exist.
On 24/01/2009, at 8:21 AM, Chris McDonald wrote:
We [AS3491] null0'd the IP earlier. Rest-of-world encouraged to do the same :/
Good luck with that. Right now they're targetting ISPrime, and you've just made the DoS even more effective for them. With any luck, the rest of the world will follow suit and the bad guys win! yay! :)
Short of getting the rest of the world to properly implement ingress filtering (ha, ha), I think dropping the specific packets that generate the reflected traffic is good enough for now. The load on the reflectors is minimal.
Nathan.
-- Mark Andrews, ISC 1 Seymour St., Dundas Valley, NSW 2117, Australia PHONE: +61 2 9871 4742 INTERNET: Mark_Andrews@isc.org