High availability secondaray DNS system

High availability secondaray DNS system

Post by Thomas Ro » Fri, 12 Feb 1999 04:00:00



Hi everybody,

Our company runs a "privat Internet" (Intranet?) distributed all over Germany;
it is not connected to *THE* Internet. We have
one primary DNS-Server and more than ten secondaries. The secondaries are
now realized by HP-clusters, formerly used to serve several services apart from
DNS.
The HPs will be removed whithin the next weeks. Thereafter, the secondary DNS
service is planned to be handled by Linux-boxes. Two Linux computers shall
be responsible for one secondary name service: one as the main
secondary DNS server
(MS), the otherone as the backup secondary (BS) DNS server
(that is what our chiefs want).

So the MS shall handle all DNS requests. If it crashes, BS takes over the
IP-adr. from
MS.

Is this possible with Eddieware? Or do I need sth. different?

Any help is highly appreciated.
Thanx and Regard,
Thomas

 
 
 

High availability secondaray DNS system

Post by Thomas Ro » Sun, 14 Feb 1999 04:00:00


: That sounds like the dilbert principle in action.  All systems I have run
: into allow you to list multiple name servers.  I would not even think about
: this switching ip address theory.  It sounds like a good way to regularly end
: up with no name services.  I would give each machine an IP address and list
: both in order in the resolver configuation of win sock setup or etc.
: Incidentally unless these machines are ojnly going to do dns I would give the
: dns its own ip address so that in a pinch you can move it fast.

The reason for the mentioned configuration is that there are several NT
PC included in the net. They read an IP for a name server during start up.
If this name server fails, the NTs are not able to switch over to another
name server unless they are rebootet. So I am desperately looking
for a software, that makes it possible to take over ip addresses from
failed servers.  

Any ideas?
Thomas

 
 
 

High availability secondaray DNS system

Post by James Youngma » Sun, 14 Feb 1999 04:00:00



> Hi everybody,

> Our company runs a "privat Internet" (Intranet?) distributed all over Germany;
> it is not connected to *THE* Internet. We have
> one primary DNS-Server and more than ten secondaries. The secondaries are
> now realized by HP-clusters, formerly used to serve several services apart from
> DNS.
> The HPs will be removed whithin the next weeks. Thereafter, the secondary DNS
> service is planned to be handled by Linux-boxes. Two Linux computers shall
> be responsible for one secondary name service: one as the main
> secondary DNS server
> (MS), the otherone as the backup secondary (BS) DNS server
> (that is what our chiefs want).

> So the MS shall handle all DNS requests. If it crashes, BS takes over the
> IP-adr. from
> MS.

> Is this possible with Eddieware? Or do I need sth. different?

No, it works out of the box anyway, because that is the way that BIND
is designed to operate.

Clients should just list all 3 DNS servers in /etc/resolv.conf.  DNS
server A is authoritative and the master.  Server B is secondary to
it.  C is also secondary to it, but it doesn't matter much if is
slaves off A or B.  Ditto for D through Z etc, if required.

--

 
 
 

High availability secondaray DNS system

Post by Bernd Eckenfel » Mon, 15 Feb 1999 04:00:00



> The reason for the mentioned configuration is that there are several NT
> PC included in the net.

NT-Servers support two nameservers. And you can change them while the systm
is running, just dont rebbot it.

Greetings
Bernd

 
 
 

1. High Availability Sun Systems

Folks,

We have requirements for a system that has to have a 98-99% availability,
and we're looking at Sun Servers and workstations.  I've read a lot about
clustering and high availability, in general, but I'm not that familiar with
the Sun line of products (just by reputation).  When we tried to get MTBF
information and the like to Sun, they said they don't hand that info out, in
general, so...

Are any of you running systems that _must_ run 24/7/365 and can not, under
any circumstances, stop passing/retrieving data?  Is the availability of one
machine (with redundant power supplies, extra cooling, hot-swappable,
RAID-equipped, etc) high enough to be trusted to support a mission-critical,
no-down-time allowed application?  Are all the machine features listed in
the previous sentence necessary? This system will mostly be un-manned.

Thanks for listening.

Angela Cole, Systems Engineer

2. Install Linux, Window95, and Window98

3. High availability system howto question

4. Short-cuts..

5. high-availability redundant server systems

6. ADSM NT using HP Surestore Optical 40st Mod.C1150A.?

7. Virtual filesystem for high availability AIX system

8. user privileges on msdos/vfat fs

9. Internet WebCast re: High Systems Availability

10. Fault Tolerance and/or High Availability systems needed.

11. Fault Tolerance and/or HIgh Availability systems needed.

12. FDDI & High availability on Linux?

13. High availability clustering