In last month's issue, I described how you can easily force all DNS queries through your own DNS servers and use RPZ (Response Policy Zones) to filter and redirect outgoing queries [1]. What about logging replies, though? Being able to block traffic based on the results returned is great, but without looking at the results, how do you know what to block?

Tcpdump, Wireshark, and Snort

Unfortunately, BIND does not have an internal logging method to log the replies of DNS queries. However, DNS is a well-understood protocol and is not encrypted (DNSSEC only signs data, it does not encrypt it), so the best method to handle logging of DNS replies is simply to run a network sniffer – such as tcpdump, Wireshark, or Snort – and to log everything on port 53 (both TCP and UDP).

I recommend logging the raw data, which has the advantage of providing data for correlation that you might not have stored otherwise (e.g., the TTL of packets could prove useful). Additionally, you can either process the network capture files in batches or add logging to a database. For example, with Wireshark's tshark utility, you can simply create a script to write the network data to a MySQL database and then pipe the network capture through the script for real-time logging to MySQL [2].

[...]