rndc flushname not working
each at isc.org
Mon Apr 13 18:45:29 UTC 2015
On Mon, Apr 13, 2015 at 11:17:41AM -0700, Frank Even wrote:
> I have to apologize for that. I'd still definitely be curious to know
> what info is stored in the ADB though since according to the docs ADB
> was never intended to be flushed with a "flushtree" (although that has
> now apparently been addressed in a new version as stated earlier in
> the thread).
The ADB is the "address database"; it holds the addresses of previously
encountered authoritative name servers and information about them such as
whether they support EDNS and the largest packet size they're known to be
able to accommodate.
The "bad cache" is a separate collection of name server addresses that have
been found to be irretrievably broken within the last several minutes. It
prevents named from wasting too much of its time trying to get answers from
(The reason flushtree didn't originally clear those, if your curiousity
extends this far, is that they're implemented as hash tables instead of
trees like the DNS cache, so deleting all records below a given name is
inefficient - it requires a linear search through the entire hash table.
It turned out to be a useful thing to do, though, so we eventually
decided to go ahead and put up with the inefficiency.)
Evan Hunt -- each at isc.org
Internet Systems Consortium, Inc.
More information about the bind-users