Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!crdgw1!ge-dab!calvin.vf.ge.com!andy From: andy@calvin.VF.GE.COM (Andy Hazeltine) Newsgroups: comp.protocols.tcp-ip.domains Subject: Re: Zone transfers for "." Keywords: zone transfers root domain Message-ID: <1991Jun21.203056.29173@ge-dab.GE.COM> Date: 21 Jun 91 20:30:56 GMT References: <1991Jun17.220952.1807@Daisy.EE.UND.AC.ZA> Sender: news@ge-dab.GE.COM (USENET News System) Reply-To: andy@ge-dab.GE.COM (Andy Hazeltine) Organization: GE Aerospace, Valley Forge, PA Lines: 27 Nntp-Posting-Host: calvin.vf.ge.com In article <1991Jun17.220952.1807@Daisy.EE.UND.AC.ZA>, barrett@Daisy.EE.UND.AC.ZA (Alan P Barrett) writes: |> In article , |> linegar@bwdls49.bnr.ca (Derick Linegar) writes: |> > In setting up a "lab" DNS system that mimics the Internet, |> > we've come to the argument where one sides insist that |> > zone transfers *do not* work for the root domain. Is this |> > true? How does the root servers keep things in sync? Wouldn't |> > it be a waste not to use zone transfers for root and use it |> > everywhere else? |> |> Among Berkeley bind 4.8.1's problems is that it makes a total mess of |> root zone transfers. I hear that bind 4.8.3 has the same problem with |> root zones, but I haven't seen the code myself. (I guess the real root |> servers don't run Berkeley bind?) |> Bind 4.8.3 does have the same problem. I have taken Alan Barrett's patches for 4.8.1 and adapted them to work for plain 4.8.3. They don't work, however, with Don Lewis's patches to detect and log lame delegations, etc. I am using bind 4.8.3 with my patches successfully to run a set of name servers for our internal network which does not directly connect to the Internet. -- Andy Hazeltine, GE Aerospace Information Technology, Valley Forge, PA andy@ge-dab.ge.com