bind8.2 slave for earlier bind + negative caching

Barry Margolin barmar at bbnplanet.com
Mon Feb 7 22:41:56 UTC 2000


In article <87nfhv$8p6$1 at nnrp1.deja.com>,  <Marc_Lampo at hotmail.com> wrote:
>if a bind8.2 server is slave of a pre bind8.2 master, is it correct to
>assume that it will not default to the 10 minutes negative caching time
>like the master?

Caching only applies to caching servers, not masters and slave servers.

>My assumption is based on the following reasoning :
>1. in a zone transfer the bind8.2 server will learn the ttl of the SOA
>record (which will be, unless explicitly specified in the zone file, the
>default minimum time (4th counter in the SOA data)
>2. I'm assuming the bind8.2 server will use that ttl when replying with
>NXDOMAIN to its queriers.  Can somebody confirm ?

It should use the minimum of the SOA record's TTL and the SOA Minimum
field.

>So, can we generalize ?  Don't make a bind8.2 server slave of a pre8.2
>version.

I haven't heard this advice.  It will be difficult to implement in
practice.  We provide slave service for hundreds of customers, and many of
them are still running BIND 4; we're certainly not going to wait for all of
them to upgrade before we upgrade to 8.2.

>Or, always put an explicit ttl on the SOA record (for the negative
>caching time) and explicit (higher) values on all other records (or at
>least the record immediately following that SOA record.  By explicitly
>specifying the ttl one doesnot rely on syntactical features that obtain
>the same result.

I don't think this is necessary.  The negative cache time is the SOA
Minimum field, not the SOA records's TTL.

-- 
Barry Margolin, barmar at bbnplanet.com
GTE Internetworking, Powered by BBN, Burlington, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.



More information about the bind-users mailing list