how to protect name servers against cache corruption

someone asked me a question in private e-mail that deserves a public answer.

1) How exactly did Eugene Kashperuff propogate this "RR poisoning" across
the Internet? From NANOG's previous mailings I can deduce that it was along
the lines of dig @victim -t ns www.alternic.net. Where www.alternic.net had
duff A records.

yes.

2) What were/are the symptoms of this attack? www.internic.net resolving to
www.alternic.net?

yes.

3) If it was that easy to do, why hasn't it happened again?

because that particular attack only works if you are willing to get caught.
since eugene did this as a publicity stunt (which, i understand, has now
begun to backfire on him since his victims didn't interpret it that way),
he _needed_ to be caught.

3a) What measures were taken (other than discussion of DNSSEC, or lack of
it) to 'cure' affected servers?

upgrade to bind-4.9.6 or bind-8.1.1.

4) How can I check for cache corruption?

"dig @0 www.netsol.com a" and "dig @cache00.ns.uu.net www.netsol.com a" and
check for differences.

Apologies if any of the above sound moronic or ill-informed; extracting
facts from reams of "what is a backhoe" mail list is a painfully slow task.
Time for some filters I think...

no apologia needed. public explainations of this attack have been poor, even
and especially by me. i'm grateful for the opportunity to improve on that.

To respond to your upgrade to bind-4.9.6 or 8.1.1, I would go with
4.9.6 because I know of many bugs left in 8.1.1 that can harm a Name
Server. When I get more information on this, I will release it to the
list.

One of the many bugs in it involves CNAME. I will talk to the person
who released this information to me and get back to the list.

> 3) If it was that easy to do, why hasn't it happened again?

because that particular attack only works if you are willing to get caught.

Nicely put. Although accidents do happen, like the genieweb.com
answering for ".com" debacle a couple weeks back.

> 4) How can I check for cache corruption?

"dig @0 www.netsol.com a" and "dig @cache00.ns.uu.net www.netsol.com a" and
check for differences.

Paul: I assume dig @0 is an idiom for localhost? (Apologies for being
less than familiar with dig, it's not on this machine, and I'm not the
admin.)

> Apologies if any of the above sound moronic or ill-informed; extracting
> facts from reams of "what is a backhoe" mail list is a painfully slow task.
> Time for some filters I think...

no apologia needed. public explainations of this attack have been poor, even
and especially by me. i'm grateful for the opportunity to improve on that.

I hadn't thought that the explanations were all _that_ weak... and I'm
on 7 lists, and the backhoe traffic didn't bother _me_ that much.

Perhaps time for a new mail program, or a faster link?

Cheers,
-- jr '30 newsgroups, too' a