I think this is the main cause of all strange things happening on the net for last few days. Vipul Rishab A. Ghosh Wrote :
Was I the only one nuked by the DNS/BIND crash yesterday? I hope I've not been automatically unsubscribed from the list. As not everyone here reads c.p.tcp-ip.d I've attached Karl Denninger's analysis. For those who were luckily immune, my ISP (best.com) like many others, had it's DNS crash for _local_ domain names (belonging to the ISP and customers like me) through most of yesterday. No, not a virus, but bad DNS records "floating around" as Karl puts it, that happened to expose a bug in the latest version of BIND.
So much for immunity to nuclear war!
Rishab
From: karl@MCS.COM (Karl Denninger) Newsgroups: comp.protocols.tcp-ip.domains Subject: SERIOUS PROBLEM WITH DNS SERVERS AND BAD RECORDS - Rev 4.9.4 Date: 23 Aug 1996 10:10:39 -0500 Organization: MCSNet Ops, Chicago, IL Message-ID: <4vkhlf$u4@Jupiter.mcs.net>
CAUTION!
There are a series of bad nameserver records floating around on the net which are blowing up BIND versions 4.9.4 (REL and T5B) and possibly other releases as well.
This has been VERIFIED to be impacting multiple ISPs and their DNS servers.
We are shutting off updates from ANY DNS server which presents bogus data, which stops it from killing our code, but is of no help to the large number of domains which are presumably rendered unreachable.
At present, this list is:
bogusns 204.94.129.65 158.43.192.7 ; bogusns 199.3.12.2 38.241.98.5 199.71.224.105 206.215.3.10 bogusns 134.75.30.253 198.41.0.4 128.63.2.53 198.41.0.4 bogusns 206.66.184.11 206.66.104.37 ; bogusns 163.173.128.6 163.173.128.254 200.6.39.1 192.33.4.12 128.174.36.254 bogusns 129.79.1.9 128.174.5.58
All of these have presented at least one malformed record to us in the last two hours!
Folks, if you run one of these servers, start tracking down the problem on your end. If this is bad cached data, THOSE AFFECTED MUST FLUSH IT AS SOON AS POSSIBLE TO TRY TO PREVENT PROPAGATION.
This problem started as an isolated set of incidents yesterday, and is now spreading like wildfire.
The actual bad data appears to be a domain name being returned in an authority record which is of the form "domain.com<tab>com". We have not yet caught a bad returned record in a debug file; that is being attempted now.
When this goes through "dn_expand" in the BIND code, it causes memory arena corruption and subsequent failure to resolve VALID zones which you are authoritative for. First signs are reports of "corrupted authority data" if you are using "dig" to check zones which you hold authority records for.
We are working on a way to "harden" the code against this kind of junk data, but until we can get one deployed our defense is to shut down communication from those who are presenting us the garbage.
PLEASE CHECK YOUR NAMESERVERS OUT AND TAKE NECESSARY STEPS YOURSELF! This is a serious problem which has the possibility of melting significant parts of the Internet infrastructure.
-- -- Karl Denninger (karl@MCS.Net)| MCSNet - The Finest Internet Connectivity http://www.mcs.net/~karl | T1 from $600 monthly; speeds to DS-3 available | 23 Chicagoland Prefixes, 13 ISDN, much more Voice: [+1 312 803-MCS1 x219]| Email to "info@mcs.net" WWW: http://www.mcs.net/ Fax: [+1 312 248-9865] | Home of Chicago's only FULL Clarinet feed!
-- bryant durrell http://www.innocence.com/~durrell durrell@innocence.com http://www.innocence.com/fengshui durrell@bofh.net http://www.innocence.com/shadowfist big black nemesis parthenogenesis no one move a muscle as the dead come home