notifies & bind 8 (extended)

Kevin Darcy kcd at daimlerchrysler.com
Wed Apr 11 18:17:11 UTC 2001


José M. Fandiño wrote:

> Hi,
>
> yes, they confirm to me that artemis do a shutdown daily (midnight) and a
> transfer update the DNS server 'artemis'. It seems a botched job, because
> I have an inconsistency time too high (up to 24h).
>
> I'm having a "rfc battle" with them, there is any document
> to advise against this unfriendly behavior??

I doubt it. The main RFC's just specify how to do zone transfers, but it is outside of their scope to specify how
quickly the zone transfers must be "committed". That would be considered an implementation detail. There are also
some "operational" RFC's, which are not considered standards, but those mostly apply to root and TLD servers, which
don't do zone transfers very frequently either.

You could look through the files in the doc/rfc directory of the BIND distribution but, as I said, I doubt you'll
find much useful "ammunition" there...

                                                                                                            - Kevin



>
>
> thanks and regards,
>
> Kevin Darcy wrote:
> >
> > Unfortunately, you'll have to talk to the maintainers of "artemis" to see why it's taking so long between the
> > completion of the zone transfer and when the new data is available from their server. Maybe they are running a
> > modified replication process in order to conserve server resources, e.g. run named-xfer manually and reload
> > once a day (midnight sounds suspicious to me), with all of the zones defined as "master" in the named.conf
> > file...
> >
> >                                                                                                             -
> > Kevin
> >
> > José M. Fandiño wrote:
> >
> > > Hi friends,
> > >
> > > I'm using a stealth master server with my ISP which have two name
> > > servers, minerva and artemis, when I do a record update and the
> > > slaves complete the notify (them give me an ACK) a server (minerva)
> > > do an AXFR seconds later and the other one have a delay of several
> > > minutes (probably as a consequence of the OS load as Kevin said).
> > >
> > > The trouble arise when after of the AXFRs I query to both name
> > > servers, minerva have a correct SOA but artemis not. If I don't
> > > understand wrong after ot the AXFRs succeed the only possible delay is
> > > the proper OS delay for write the transfers to the zone files, but
> > > I get up to 16 hours!!! of delay.
> > > Close to midnight artemis seems answer correctly the SOA for my zone.
> > >
> > > I guess that they have artemis in any type of read-only mode
> > > and a cron job do a ftp/ssh/rsync/... transfer. This is any
> > > shadow security trick? :-? ¿human stupidity? :)
> > > Don't break this the DNS consistency?
> > >
> > > the worst is that is people manage the 45% of the zones under
> > > the .es TLD.
> --
> -----BEGIN GEEK CODE BLOCK-----
> Version: 3.1
> GCS d- s+: a- C+++ UL++++$ P+ L+++ E--- W++ N+ o K- w---
> O+ M+ V- PS PE+ Y PGP+>+++ t+ 5 X+++ R- tv@ b+++ DI-- D+++
> G e- h++ !r !z
> ------END GEEK CODE BLOCK------





More information about the bind-users mailing list