Refreshing cache in other DNS servers

[/dev/rob0]

On Tue, Oct 15, 2013 at 10:53:13PM +0800, babu dheen wrote:
> Hi Matus,
> "The standard way to handle this situation is, when you know you 
> are going to make a change, to lower TTL of a particular RR to a 
> small value (e.g.  300) and after change to restore the TTL to
> sane standard value (e.g.  43200)."
>  
> I just need clarification on your above update.
>  
> If I change the TTL value on the particular zone after modifying
> a record in Redhat Linux BIND Caching DNS server, My Redhat bind 
> Caching DNS server cache would be refreshed after 300 seconds but 
> what if my backend windows  DNS server is still responding to  end 
> user old record from  from its cache?

A standards-compliant nameserver caches a record exactly as long as 
the TTL. Something queries a name/class/RRtype, for example:
"walk-on.us/IN/A", and this is the answer:

walk-on.us.             7200    IN      A       66.226.201.55

That's two hours. If another client of this caching nameserver asks 
for that same record one second later, it would get it with TTL of 
7199. Every second that TTL value decrements by one. If a client 
queries that record 7201 seconds later, the caching nameserver must 
retrieve it again from the authoritative server[s] for that zone.

If your Windows DNS server is not standards-compliant all bets are 
off, and you will have to ask the vendor for support. This is the 
BIND-users list.

> So my backend windows DNS server can get the newly modified record 
> from DNS only when its contacting Redhat DNS server for the newly 
> added date once Windows DNS cache is refreshed?

There too, you are asking us to support your Windows DNS server. I 
can suggest this: if your Windows DNS server is not standards- 
compliant or if you cannot get support for it, you might consider 
replacing it with the Windows port of BIND.

Good luck.
-- 
  http://rob0.nodns4.us/ -- system administration and consulting
  Offlist GMX mail is seen only if "/dev/rob0" is in the Subject: