[Dnsmon-test] retries of DNS queries
Chris Amin
camin at ripe.net
Fri Feb 14 15:01:30 CET 2014
On 13/02/2014 17:29, Gilles Massen wrote:
> Hi,
>
> On 02/13/2014 05:16 PM, Chris Amin wrote:
>
>> In addition, we have changed a line in the testers page in the section
>> comparing the old and the new DNSMON implementations. We previously
>> stated that the RIPE Atlas measurements are not retried on failure
>> (whereas the old TTM boxes do carry out retries). However, RIPE Atlas
>> has gained the ability to carry out an arbitrary number of retries, and
>> uses a value of 10 by default. The data that we have collected so far
>> for the new DNSMON, and the visualisations, are therefore based on this
>> methodology.
>>
>> We don't want to immediately change to using 0 retries, nor do we want
>> to change to match the exact settings of the old DNSMON, without
>> gathering feedback from you. So if you have a strong opinion on this,
>> please let us know.
>
> No strong opinion (yet), but a request for clarification: how do the
> really late answers relate to retries? i.e. is a query is answered after
> 3s, does that mean that
>
> a) an answer came in 3s after the first query and after an unknown
> number of retries
>
> or
>
> b) the answer came 3s after the exact packet that asked the question?
>
> (if retries use the same query id / source port the answer is probably
> obvious...)
Hi Gilles,
The RT value includes just the successful (i.e. the last) request.
Regards,
Chris
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2466 bytes
Desc: S/MIME Cryptographic Signature
Url : https://www.ripe.net/mailman/private/dnsmon-test/attachments/20140214/d23e5147/attachment.bin