bind9 answers no queries during AXFR
Chris Buxton
cbuxton at mac.com
Tue Aug 10 20:38:21 UTC 2004
The ideal solution would be to convince whoever owns the master server
to enable dynamic updates, and then to use them to update the zone.
That way, your server requests an IXFR instead of an AXFR, and the
whole thing goes a lot faster. But this may require them to rewrite any
automated system that modifies the zone.
If that's not an option, you could in theory recreate the axfr process.
Set your server up as a master, using the zone file already on disk.
Then write a script to poll their server for updates, perhaps once an
hour, and get a zone transfer (using 'dig' redirected to a file). Then
tell your server to reload the zone from disk. This should, in theory,
lead to less downtime than acting as a slave, since disk access is so
much faster than network access. But there will still probably be a
noticeable hiccup in the operation.
Regards,
Chris
On Aug 9, 2004, at 7:46 AM, Martin H. Sluka wrote:
> We got a BIND9 server which is a secondary for a very large DNS
> zone (an RBL; about three million entries, 70 MB as file, several
> hundred MB in RAM) that usually updates each few hours. Even on
> current hardware, the zone transfer (or reading the zone after
> the transfer) takes several minutes.
>
> The problem is, that BIND9 does not seem to answer _any_ queries
> during the update phase. Any help to solve this problem would be
> appreciated. (Splitting the zone is, however, not possible.)
>
> Regards,
> Martin
>
More information about the bind-users
mailing list