changing ttl of mx record

Kevin Darcy kcd at chrysler.com
Mon Oct 10 17:51:57 UTC 2011


On 10/10/2011 11:13 AM, enigmedia wrote:
> On 10/10/2011 9:26 AM, Albert E. Whale, CHS CISA CISSP wrote:
>> If you are going to update the IP and TTL, why not adjust both?  This will
> take care of some broken DNS packages.
>> Hth
> Thanks, I had googled around a bit and saw some conflicting opinions about
> whether TTL'ing the MX was necessary or even a good idea...but I don't know if
> that concern is still true or not?
>
>> Sent from my iPhone
>>
>> On Oct 10, 2011, at 10:42 AM, "enigmedia"<online-reg at enigmedia.com>  wrote:
>>
>>> Hi All: If I need to set a short TTL prior to an MX IP change, do I need to
>>> modify the TTL of the MX record, or just the A record the MX points to?
>>> (There's just a single A record for the MX).
>>>
>>>
>>> _______________________________________________
There's no point ever changing the TTL of records that are not going to 
change. All that does is increase traffic unnecessarily.

So, if the MX record stays the same, but the A record(s) to which the MX 
target(s) resolve are going to be changing, modify the TTL of the A 
record(s) only.

You should only change both if both sets of records are going to change.

I'm not sure what "brokenness" is being referred to. Is some 
implementation of a DNS resolver going to *stop* resolving an MX record 
just because the TTL of the A record(s) of the MX target(s) expired and 
were re-fetched? I've been doing DNS and SMTP for a _long_ time, and 
I've never seen such "brokenness"...

                                                                         
                                                                         
     - Kevin




More information about the bind-users mailing list