From Darren.Gamble at sjrb.ca Tue Jun 2 12:38:37 2009 From: Darren.Gamble at sjrb.ca (Darren Gamble) Date: Tue Jun 2 19:39:13 2009 Subject: [Pdns-dev] Possible bug in authoritative server CNAME toremote domains? In-Reply-To: <20090526133738.GA10135@utemp.mieszkanie> References: <20090526133738.GA10135@utemp.mieszkanie> Message-ID: <00F10C2766A9664D8E1BE7D147015B0206E377E4@PRDCG4EXVW01-3.OSS.PRD> > Please look at: > http://mailman.powerdns.com/pipermail/pdns-users/2007-August/004666.html > http://mailman.powerdns.com/pipermail/pdns-users/2007-August/004667.html > > Your PowerDNS server is authoritative to wfdnstestscript.me but not to > blogware.com, is it? > > > > So it seems that even pdns comments thinks it shouldn't be sending the > > SERVFAIL response in the external CNAME redirection case, however it is. > > Only if recursion is enabled. Maybe there is something that I am missing here, but I'd otherwise have to respectively disagree. You can serve an out-of-balliwick CNAME data for an in-balliwick name. If recursion is disabled, the server won't do any work for you, but NOERROR should be returned. The cache won't believe anything you say about that name, of course, but should still independently chase down the applicable RR of the target. To my knowledge all DNS caches do this, including pdns-recursor (yes, I did just test it now to be sure). We do also have this setup for a handful of our own names, although we don't use the auth PowerDNS server here, just the recursor. It works fine - exactly as Mark Zealey is expecting it to. Perhaps Bert can comment? ============================ Darren Gamble Systems Architect, Regional Services Shaw Cablesystems GP 630 - 3rd Avenue SW Calgary, Alberta, Canada T2P 4L4 (403) 781-4948 From powerdns at usenet-verwaltung.de Fri Jun 26 13:52:19 2009 From: powerdns at usenet-verwaltung.de (Juergen Ilse) Date: Fri, 26 Jun 2009 13:52:19 +0200 Subject: [Pdns-dev] Powerdns rev1361 and newer on Big Endian Machines Message-ID: <20090626115219.GA11296@usenet-verwaltung.de> Hello, I tried to use pdns_server (rev1363, rev1365, rev1366) on a Sparc machine with Solaris8 (compiled with gcc-3.4.6 using mysql backend and pdns_recursor 3.1.7 as recursing backend). I noticed, that this combination did *not* work: i got error-messages like Jun 18 10:47:29 ns6 pdns[13946]: [ID 702911 local0.error] Discarding untracked p acket from recursor backend with id 27240. Contrack table size=3 which disappear when i comment out (or remove) the lines between #ifdef WORDS_BIGENDIAN #endif in pdns/dnsproxy.cc (it is line 174 in rev1366). Maybe these lines are a relict of older sources and are not needed any more. rev1341 worked *with* these lines in dnsproxy.cc, rev1366 seem to work only *without* these lines in dnsproxy.cc on Big endian machines (like Sparc) ... I have not yet searched, wich changes in the source made these lines unnecessary, but i think it is a bug in these newer releases in svn, which needs to be corrected (it only appears on Big Endian machines, because on other machines these line were not compiled because WORDS_BIGENDIAN will not be defined ...). I would be glad, if you fix these issue in the current svn source. Thank you. bye, Juergen Ilse -- Ein Domainname (auch wenn er Teil einer Mailadresse ist) ist nur ein Name, nicht mehr und nicht weniger ...