bind-users Digest, Vol 2230, Issue 1

[Grant Taylor]

On 10/20/2015 08:49 AM, Harshith Mulky wrote:
> We have an implementation where, once the DNS Servers are down, The 
> Client (Our device) Blacklists the IP address of DNS Servers for some 
> period of Time

How long is "some period of time"?  Is it something reasonable (read:
minutes) or something obscene (read: WAY TOO LONG)?

Does your client (device) support falling back to secondary name servers
in a sane manner?

> It can only whitelist the server when it receives periodic Responses to 
> a NAPTR Request.

It sounds like you have part of your answer to "what kind of messages
(queries)..."

> What I did find was even though Our Client was able to send periodic 
> NAPTR requests, we are unable to check what kind of NAPTR requests are 
> sent out

I would suggest enabling query logging, especially if you can properly
target your client (device). At least for diagnostic purposes.

I would also seriously consider a packet capture.  (I think pcapcs are
easy and faster than query logging, especially for ad-hoc things like this.)

> Hence my question,
> What Kind of messages are required by the client to be sent towards 
> server to determine if the DNS IP is reachable or not?

That is an application question, not a DNS server question.  As others
have stated, your application support is going to be best qualified to
answer that question.  Short of that level of help you can try to
reverse engineer it yourself via query logs and / or packet captures.



-- 
Grant. . . .
unix || die