On Tuesday July 10, 2008 a number of vendors including Microsoft, Cisco, Juniper and RedHat released patches and/or acknowledged the flaw existed. The Internet Software Consortium, the group responsible for development of the popular Berkeley Internet Domain Named (BIND) server from which nearly all DNS offshoots are based, also acknowledged the flaw and released a patch.
I personally spent about 90 minutes on last Wednesday updating several internal and external systems including numerous CentOS v5.2 servers and Windows 2003 Service Pack 2 servers. I was unable to find any mention of the DNS flaw on the Alcatel-Lucent website so I’ll probably need to place a call concerning Alcaltel-Lucent’s VitalQIP product.
I used yum to patch the CentOS Linux servers [“yum update”] and then just restarted the named process [“service named restart”]. On the Windows 2003 Service Pack 2 servers I used Windows Update to download and install KB941672 after which I rebooted the servers.
Here are some references:
I would strongly suggest that all network administrators start looking into patching their DNS servers as soon as possible.
UPDATE: July 14, 2008
Here’s an update from RedHat concerning the configuration (named.conf) of BIND;
We have updated the Enterprise Linux 5 packages in this advisory. The default and sample caching-nameserver configuration files have been updated so that they do not specify a fixed query-source port. Administrators wishing to take advantage of randomized UDP source ports should check their configuration file to ensure they have not specified fixed query-source ports.
It seems that a check of the configuration file would be in order. Let me throw in a quick warning though if your DNS server is sitting behind a firewall you may need to check with the firewall administrator to understand how the firewall will behave if you randomize your source ports. I believe there are quite a few firewalls out there that only expect to see DNS traffic sourced from a DNS server on UDP/53.