[dns-operations] .MW inconsistent zone updates?

[Stephane Bortzmeyer]

On Thu, Jun 25, 2015 at 10:23:46AM +0200,
 Gunter Grodotzki <gunter at grodotzki.co.za> wrote 
 a message of 47 lines which said:

> I did a domain update last week on cheki.mw, but it seems like some
> OPs are either sleeping or their syncing is not really working ;)

Inconsistencies are always fun to observe (remember the DNS is only
loosely consistent). For instance, 14 % of RIPE Atlas probes see the
late server. If the Atlas probes are representative (which is highly
doubtful), it means 14 % of the users will go to the old name servers:

% python resolve-name.py -t NS cheki.mw 
Measurement #2057637 for cheki.mw/NS uses 499 probes
[] : 4 occurrences
[ns-1022.awsdns-63.net. ns-1137.awsdns-14.org. ns-1722.awsdns-23.co.uk. ns-279.awsdns-34.com.] : 70 occurrences
[arch.ns.cloudflare.com. athena.ns.cloudflare.com.] : 416 occurrences
Test done at 2015-06-25T09:13:14Z

> Others, like the following, show the correct entry:

One very big cloud US hoster or the other one, does it matter? :-)