brunner@bullhead.uucp (07/24/90)
In article <23924@uflorida.cis.ufl.EDU> seeger@bikini.cis.ufl.edu (F. L. Charles Seeger III) writes: >In article <1990Jul19.233916.14458@ibmpa> brunner@ibmsupt.UUCP () writes: >| (a reply to a netnews query about hc from a user in Israel, including a mail reply to a similar query sent from a user in the Netherlands, with two silly errors. Omitted to save space and face.) >Everyone's name resolution might work better if the correct name of our >machine is used: "bikini.cis.ufl.edu" (that is "cis", not "cs"). This >machine is also reachable as "cis.ufl.edu" and "ufl.edu". It is currently >running with two Ether interfaces with IP numbers of 128.227.224.1 and >128.227.128.2. Also, for what it is worth, this naming is done with A RRs, >not CNAME RRs. Thanks for the corrections Chuck! Both A records and CNAME records are named resource records. CNAME is the canonical name for an alias, e.g., the host gar.paloalto.ibm.com can also be reached using the name dallas.paloalto.ibm.com (locally to the paloalto.ibm.com domain, only the host names need be used), e.g., in /etc/named.boot: primary paloalto.ibm.com /etc/named.hosts The primay (authoratative) data for the paloalto domain is in /etc/named.hosts (the reverse form for IN-ADDR lookups is in /etc/named.rev), there gar has a CNAME of dallas: gar IN A 9.49.22.13 dallas IN CNAME gar In a perfect world, ibmpa.paloalto.ibm.com (the host) would have a CNAME of paloalto.ibm.com, so that telneting to the domain would be equivalent to telneting to a well-known host in that domain, and sending mail to user@domain would have the affect of sending mail to user@mailhost. In the same primary data file, a host may be specified with multiple A RRs, (A resource records are the ip address aka "dotted quad" which corresponds domain name), e.g., in /etc/hamed.hosts, gilroy IN A 9.49.19.19 sashimi IN A 9.49.19.19 Or at ufl.edu one_name IN A 128.227.224.1 other_name IN A 128.227.128.2 This "dual name" convention local to paloalto allows hosts to have a "fish" name when running IBM/4.3, and a "town" name when running AIXx.y. The more usual (and intended form) is used at ufl.edu, where multiple network addresses are matched to possibly different hostnames, which is a single actual host. Personally I prefer the CNAME approach for the mail user@domain capability, where a single actual mailhost (which may not be up when mail delivery is attempted) is replaced by a logical mailhost, which is which ever of the multiple nameservers (primary, secondary, slave) is up at the moment, with local delivery effectively queued in the named/sendmail interaction until the final host (or domain) is up. >Thanks again to Eric and Mark for their continuing support of the "good" >brand of IBM *nix. And thanks for the thanks, and thanks for the archive! > Now, if we could only get a 4.3/4.4 port to the RS6000... See your sales rep regarding requirements for software on the RIOS. IBM wants to know your thoughts on AIX3.1 and the RIOS hardware. Keep your comments as constructive and computer-ish as possible, as flames buy nothing. Eric Brunner, Consultant, IBM AWD Palo Alto (415) 855-4486 inet: brunner@monet.berkeley.edu uucp: uunet!ibmsupt!brunner trying to understand multiprocessing is like having bees live inside your head.
seeger@bikini.cis.ufl.edu (F. L. Charles Seeger III) (07/25/90)
I didn't mean to start a name server discussion here. Sorry 'bout that. In article <1990Jul23.185122.7540@ibmpa> brunner@ibmsupt.UUCP () writes: |Personally I prefer the CNAME approach for the mail user@domain capability, |where a single actual mailhost (which may not be up when mail delivery is |attempted) is replaced by a logical mailhost, which is which ever of the |multiple nameservers (primary, secondary, slave) is up at the moment, with |local delivery effectively queued in the named/sendmail interaction until |the final host (or domain) is up. Your example isn't real clear to me, but it seems that MX RRs would be more appropriate than CNAMEs for this mail scenario. In our case, we wish things other than mail (including non-MX MTAs) to work with "ufl.edu". Using an A RR rather than a CNAME puts less burden on the name servers and should be more robust. For "less important" names the CNAME approach might be preferable for easier maintenance. BTW, "ufl.edu" and "cis.ufl.edu" are in different zones of authority, and we stick in the A RR as part of the SOA RR. If anyone wants to continue this discussion, we should take it elsewhere. |And thanks for the thanks, and thanks for the archive! You're most welcome. |> Now, if we could only get a 4.3/4.4 port to the RS6000... | |See your sales rep regarding requirements for software on the RIOS. IBM wants |to know your thoughts on AIX3.1 and the RIOS hardware. Keep your comments as |constructive and computer-ish as possible, as flames buy nothing. Always. Anybody written a fstab to filesystems translator script that they wouldn't mind sharing? (Yes, I'll grovel to save an hour). Regards, Chuck -- Charles Seeger E301 CSE Building Office: +1 904 392 1508 CIS Department University of Florida Fax: +1 904 392 1220 seeger@ufl.edu Gainesville, FL 32611-2024 Home: +1 904 375 1819