Source code question

[Kevin Darcy]

This is more than a "little" hack, since the in-core database has all sorts of
internal references (e.g. pointers) which would need to be "flattened" when
writing the data out to non-volatile storage. Also, you'd need to timestamp
each record so that stale entries could be purged when the database is
reloaded. Not to mention that the whole process of reloading the cache,
converting it to internal form, and purging the stale entries, whenever named
restarts, is likely to have a worse impact on query latency than just
regenerating the cache from scratch.

BIND 9 has support for database backends, but I'm not sure whether you can have
a database backend for cache data -- mostly people are interested in putting
authoritative data into a database, not cached data.


- Kevin

Carsten Muck wrote:

> Hi,
>
> I´m running Bind8 on Suse Linux 7.1,
> here is my question:
>
> Can anyone tell me where and how Bind stores the cached data in the Memory?
>
> We are a group of 5 students getting rid of loosing all the cached data when
> restarting named.
>
> So we want to write a "little" Hack or Patch to make it possible to store
> the data in a file before shutting down the nameser, and reload the data on
> demand.
>
> The worst problem is to find the regarding part of the code in the src of
> bind (does anyone know where the arrays or structs for the data are, or
> where the array is filled with the data?)
>
> Regards
>
> Carsten Muck - www.it-einstieg.de