[rbldnsd] Warning: possible danger of using rbldnsd, and upcoming data format change

Jon Lewis jlewis at lewis.org
Tue Jun 8 06:21:19 MSD 2004


On Tue, 8 Jun 2004, Michael Tokarev wrote:

> In order to reduce a risk to list larger netranges due to corrupted
> input, I plan to modify rbldnsd in such a way so it will reject
> incomplete IPv4 addreses, disallowing prefixes.  In order to specify the
> /8 listing for example, one will have to type 3 zero octets, like
> 127.0.0.0/8, not like 127 or 127/8.  Well... maybe the 127/8 form still
> will be accepted.

What about things like:

4.192-218

In order to save on bytes but also make it clear an IP range/entry is
complete, how about requiring that it be a full 4-bytes dotted quad or be
: terminated?  Thus the above could be entered as:

4.192-218:

Would that work?

> Another change I'm planning is to constrain the
> minimum prefix length (per dataset), and to reject
> the whole dataset in case of too many parsing
> errors.  But I have to think about the features
> first... ;)

You just keep making the 0/0 list harder to do :)

----------------------------------------------------------------------
 Jon Lewis                   |  I route
 Senior Network Engineer     |  therefore you are
 Atlantic Net                |
_________ http://www.lewis.org/~jlewis/pgp for PGP public key_________


More information about the rbldnsd mailing list