On Fri, Nov 04, 2005 at 12:55:57PM -0600, Les Mikesell wrote: > > But this can also be accomplished by using an A record that points to > > 10.0.0.1 instead of the CNAME... > > Then you have the issue of reverse lookups which have to > resolve to only one of the names. Nothing enforces it, but in > theory the name with the A record should be the same as the > reverse lookup and the others should be CNAME aliases. I do this all the time; it works just fine if you use A records. Most software that checks reverse lookups normally only checks that the name resolves, not that the forward and reverse lookups match. However Sendmail can be configured to do that check, and probably other server software can as well, but since in practice this mismatch happens so often on the Internet, it's basically never a good idea to turn those features on. > The only quirk this introduces is that the CNAME entry may have a > different time-to-live than the A record. Of course, all of your A records for the same IP can have different TTLs as well, so that's really not very quirky. Every record can have its own. It's generally a bad idea to do that though, because the inconsistency increases the likelihood of making mistakes. > myname IN CNAME yourname.yourdomain.com. > (and by the way, the CNAMEs above are incorrect - they should use > the full domain name with a trailing dot as the target even when > in the same domain - if the bare names work it is an accident due > to the clients adding a default during lookup.) Er, no... that's how named zone files are interpreted. If you don't include a trailing dot, BIND automatically adds the zone name (as determined in named.conf) to the end of it. It's not an accident. There's nothing special about CNAME records in that regard. -- Derek D. Martin http://www.pizzashack.org/ GPG Key ID: 0x81CFE75D
Attachment:
pgpgmHGwPjDXZ.pgp
Description: PGP signature
