DNS Server service

  • Thread starter Thread starter dsmcd
  • Start date Start date
D

dsmcd

Hello...

Three times now, every three days I've gotten calls saying 'internet is down'.
I can ping anywhere on the lan and they can do email, so they have 'internet',
but they can not browse the internet. All clients have their primary dns
pointing to this fileserver.

Restarting the DSN Server server does the trick, but I'm wondering what may be
causing this. All event logs, including the DNS Server logs are clean. The DNS
Server logs only record the stoppping and starting of the service. The DNS
system has been working properly with no changes, for a few years.

Whether it's a coincidence that this has happened once every three days three
times in a row, I don't know.

Any suggestions?

Thx,
D.
 
Now you got me wondering, I have been having similar odd things with
programs that have hitherto been totally reliable both at home and at work
all day today and yesterday.

Example WeatherWatcher which connect (I think) to Weather.com periodically
has been saying "Cannot find internet" yet mail worked at the same time...
similar with our local Doppler radar link... that's been sporadically coming
up "Page not found".

Maybe there is some global DNS trouble?

Charlie
 
Charlie Tame said:
Now you got me wondering, I have been having similar odd things with
programs that have hitherto been totally reliable both at home and at work
all day today and yesterday.
Click to expand...

Ok guys, Here is the 85% likely cause of these problems:

All internal clients must point at STRICTLY the
INTERNAL DNS server (set.)

What you have likely done is point the clients PREFERRED
and ALTERNATE to DIFFERENT sets of servers.

(Clients don't know about Primary and Secondary DNS --
that is server side terminology)

It's a common mistake to think you can get them to fail over
that way, but it is never reliable and leads to spooky problem
reports.

Clients "latch" on to one server, but if they can't find that
one 'in time' they latch onto the other -- and might switch
back after a while but not immediatly when the first becomes
available.

Point all internal clients to ONLY the internal DNS -- have
the internal DNS forward to the external DNS in your DMZ
or at the ISP.

Remember that "servers" are DNS clients too -- especially
in Windows AD domains. So this same rule applies to your
DNS servers, DCs and other servers.

Point NIC->IP Properties-> DNS to strictly the internal DNS.
 
Herb Martin said:
Point NIC->IP Properties-> DNS to strictly the internal DNS.
Click to expand...

Yup. That what the clients have. I said 'primary' when I meant 'preferred'.
They have no secondary.

Thanks again.

Any other suggestions?

Thx,
D.
 
dsmcd said:
Yup. That what the clients have. I said 'primary' when I meant 'preferred'.
They have no secondary.
Click to expand...

Only one?

Only set to Internal DNS?

(Well, that ruins THAT theory <grin>)

Ok, check the forwarders on the DNS server(s),
make sure they are focused only on EXTERNAL
Servers -- you don't want a forwarding loop
from internal Server A to internal server B
back to A etc.

Client -> Internal Server -> DMZ or ISP DNS
 
Herb Martin said:
Only one? Yup

Only set to Internal DNS? Yup

Ok, check the forwarders on the DNS server(s),
make sure they are focused only on EXTERNAL
Servers
Click to expand...

Yup. Forwarders enabled, two 'external' IPs verified with the ISP, time-out = 5
secs, recursion = unchecked.
Client -> Internal Server -> DMZ or ISP DNS.
Click to expand...

Yup.

Got a recommendation from elsewhere to set the DSN Server service's Recovery
options to restart the service on failure, which may trigger an event log if
this happens again. Barring that, it was suggested to restart the server which
I'd rather not do right now as that may prevent me from understanding what's
happening (Hey! If I kick it works again, whaddaya know!). Other than that the
only other suggestion so far has been to set the client's DNS to another DNS
server and see what happens. I'll hold off on that too since that's pretty time
consuming.

Thanks for you input. Appreciate it,
D.
 
dsmcd said:
Yup. Forwarders enabled, two 'external' IPs verified with the ISP, time-out = 5
secs, recursion = unchecked.


Yup.

Got a recommendation from elsewhere to set the DSN Server service's Recovery
options to restart the service on failure, which may trigger an event log if
this happens again.
Click to expand...

Sorry I missed the server was actually Dying. By reading
another response which looked like the normal intermittent
case.

Yes, do the restart thing and check even logs, but also consider
doing a repair install.

And then checking windows update.

DNS servers don't (often) quit for no reason.

Repair install: Original CD, install, same directory, MAKE
SURE it ask if you want to repair and choose that.

Barring that, it was suggested to restart the server which
 
What you have likely done is point the clients PREFERRED
and ALTERNATE to DIFFERENT sets of servers.

(Clients don't know about Primary and Secondary DNS --
that is server side terminology)

It's a common mistake to think you can get them to fail over
that way, but it is never reliable and leads to spooky problem
reports.

Clients "latch" on to one server, but if they can't find that
one 'in time' they latch onto the other -- and might switch
back after a while but not immediatly when the first becomes
available.
Click to expand...

Actually the problem seems to have cleared up today Herb but nevertheless
you did add something which makes a lot of sense and I guess depending on
loading (at least on the slow thing at work) it might have been at least
half the cause.

Charlie
 
Back
Top