Could anyone please answer the following for me:
If a DNS server requests the record for a domain, but the primary server is
unavailable and the record is retrieved from the secondary DNS server, how
log will the DNS server continue using secondary server over the primary?
Any Help would be gratefully received
Thank You
Lumux
> -----Original Message-----
> From: bind-use...@isc.org [mailto:bind-use...@isc.org] On
> Behalf Of lumux
> Sent: Wednesday, January 15, 2003 6:31 AM
> To: comp-protoc...@isc.org
> Subject: Secondary DNS Servers
>
> Hi All,
>
> Could anyone please answer the following for me:
>
> If a DNS server requests the record for a domain, but the primary
server
> is
> unavailable and the record is retrieved from the secondary DNS server,
how
> log will the DNS server continue using secondary server over the
primary?
It is important to remember that Primary and Secondary (now Master /
Slave) do not carry any priority. Therefore, a name server may in fact
decide to use the slave server for a zone if it responds faster than the
master.
BIND uses an algorithm to pick to best name server to use for the
resolution of a particular domain name. (someone else will pipe in with
the particulars I am sure, or, you can search the archives as I know it
has been described before here on the list) BIND basically queries all
the listed name servers. The one that responds first wins. If that
name server suddenly stops responding the process is started again until
another name server wins. I do not recall the conditions under which
the faulty name server will begin to be used again after it comes back
on line...
Dave...
>Hi All,
>
>Could anyone please answer the following for me:
>
>If a DNS server requests the record for a domain, but the primary server is
>unavailable and the record is retrieved from the secondary DNS server, how
>log will the DNS server continue using secondary server over the primary?
>
I'm not sure exactly how you're using the term "primary" and
"secondary". Order is not significant amongst the nameservers in an NS
RRset, but most nameserver implementations will prefer *faster* NS'es
over slower ones. So if one of the NS'es, which is normally faster, goes
unavailable, then the time it takes for it to be preferred again after
being restored depends on how *much* faster it is and how often it
happens to be queried from any other given nameserver -- they all have
to "re-learn" about the response speed of the restored nameserver.
- Kevin
> Could anyone please answer the following for me:
> If a DNS server requests the record for a domain, but the primary server=
is
> unavailable and the record is retrieved from the secondary DNS server, h=
ow
> log will the DNS server continue using secondary server over the primary=
?
> Any Help would be gratefully received
Nameservers will generally keep track of all nameservers for=20
a given domain, keeping statistics about response time.
The algorithm is such that all servers will be tried, but the
"best" one will be used most of the time. Eventually the dead one
will get a query, and any reachability change will update=20
reachability figures.
I think it's called "round trip time" evaluation.
> Thank You
> Lumux
--=20
Peter H=E5kanson=20=20=20=20=20=20=20=20=20
IPSec Sverige ( At Gothenburg Riverside )
Sorry about my e-mail address, but i'm trying to keep spam out,
remove "icke-reklam" if you feel for mailing me. Thanx.