Change client-side DNS cache time out value

scottyp55 · 10-23-2009, 01:32 PM

Hello,

My problem is that I have 2 domain controllers/DNS servers and when one goes down, clients hang on to the DNS cache for 1 day (by default on Windows) so they don't automatically find the other DNS server. I was able to find this setting on Windows machines in the registry and change it to a lower value, but i also have Linux servers using these for DNS. Does anyone know how I can make these changes on Red Hat Linux?

Thanks,
Scott

MensaWater · 10-23-2009, 01:37 PM

You would set it in /etc/resolv.conf.

From "man resolv.conf" on CentOS 5:

Quote:

timeout:n
sets the amount of time the resolver will wait for a
response from a remote name server before retrying the
query via a different name server. Measured in seconds,
the default is RES_TIMEOUT (currently 5, see <resolv.h>).

scottyp55 · 10-23-2009, 04:05 PM

Thanks. I'll give that a shot.
Is the n value seconds? The default of 5 seems quick, considering Microsoft's is 24 hrs.

MensaWater · 10-26-2009, 12:09 PM

It is seconds.

24 hours seems extremely long. If you had one DNS server down would you really want to wait a whole day before another DNS server responded?

I suspect you're confusing TTL (time to live) settings with client side lookup. The client uses what is called a stub resolver. DNS servers often cache records as well. The TTL deals with how long to cache a record before asking again. Windows clients (stub resolvers) do cache entries however, UNIX/LINUX clients don't typically cache - they always ask the question of the name server which itself might have cached the answer.

So the TTL deals with how long to cache an answer once you've gotten it.

The timeout I mentioned in my prior response deals with how long to wait until you ask a different server which isn't the same thing.