Newsgroups: comp.protocols.tcp-ip.domainsFrom: marka@syd.dms.CSIRO.AU (Mark Andrews)Subject: Re: BIND 4.9 bug?: Losing RRs at zone boundriesMessage-ID: <C95n1r.G37@syd.dms.CSIRO.AU>Sender: news@syd.dms.CSIRO.AUOrganization: CSIRO Division of Mathematics and Statistics, AustraliaReferences: <1993Jun24.191743.23086@cs.cornell.edu>Date: Fri, 25 Jun 1993 02:03:27 GMTIn article <1993Jun24.191743.23086@cs.cornell.edu> parmelee@cs.cornell.edu (Larry Parmelee) writes:>We've just observed a problem today where our SECONDARY nameservers>lost some critical MX and A records.  The RRs in question were MX and A>records for some of our subdomains, for example "CS.CORNELL.EDU" lost>Anyone else seen this problem?  Anyone have a fix?>>-Larry Parmelee>parmelee@cs.cornell.eduThis is a side effect of the switch to 4.9. Older buggier nameserverspassed out MX and A records for the child zone. When the parent zone isupdated it no longer has theses bogus records transmitted and thesecondaries dutifully note that this has happend and cease to know aboutthen.This can be fixed by	1. restart all the secondaries (the cached zones are ok) or	2. update the serial numbers of all the child zones.As far as I can tell it only happens when a nameserver is secondaringboth the parent and child zone.When a primary nameserver switches to 4.9 I recommend updating allserial numbers for the zone that it is a primary, and for all zones thatare children of those it in primary for.Mark.