Hi, for the last year or so im having a similar problem from the one described here but from a different perspective:
In short, i have two dynamic records (@ and www subdomains)
Every 5 months or so when the ip needs to be changed i send two GET requests (using ssl, but i tested this without ssl too)
I Made a simple script to test this behaviour using httest (
http://htt.sourceforge.net/cgi-bin/cwiki/bin/public)
Script:
http://pastebin.com/CNnLHV0J
NOTE: just to keep the problem clean, the problem still occurs using software clients for this purpose. (Directupdate, ddclient, etc)
Host tested:
dynamic.zoneedit.com and api.cp.zoneedit.com
Files tested:
dyn/tomato.php, /dyn/generic.php, /auth/dynamic.html, and many others.
Params used:
it depends, for the php files for example, ?hostname=$host1&myip=$ip
for html files, ?host=$host1&dnsto=$ip
I even tryed only ?hostname=$host1 so to set the origin ip automatically.
They all work correctly..
Exept when "sometimes" when two request are sent at the same moment to different sudomains (in this case
www.domain.com and domain.com), only the first one sent (tested this with sniffer in hand too) gets updated.
Both requests get the response <<SUCCESS CODE="200" TEXT="*ZONE* updated to *IP*" ZONE="*ZONE*"> OK
Doing a dig test on dns1, dns2, ns7, ns14, dns0 etc, and even on zoneedit.com website panel i see that one of the records (the last updated) is not updated even when i got OK response.
So i began digging in (no pun intended) and saw that this problem seems erratic, i mean, sometimes both get updated and sometimes only the first sent (within milliseconds of difference)
The other thing is that the TTL record for dynamic records get set to 0 (zero) even when changing it manually it gets reseted.
All these test where made with many variations, even switching to http 1.0, i mean, everthing i could think off. my conclusion is that zoneedit discards an update if its zone is updated too fast maybe depending on the current server load.
Other weirdness i encountered (rare) are SOA serial numbers that are way ahead of the primary server and stay that way for a long time.
clarification on the last comment: dns0.zoneedit.com is set as my primary nameserver, xfr0.zoneedit.com as the master transfer server. then i have the NS dns1.zoneedit.com NS ns7.zoneedit.com NS ns14.zoneedit.com and other remote NS on other servers. All had the same serial number exept ns7.zoneedit.com with a number way ahead of the rest, i waited for them to sync internally on zoneedit, like, 1 hour, and it stayed the same.
So my conclusion is that there is some problem on the zoneedit side when a record gets updated too fast.
My solution was to delay the update of www A record in this case and send the @ A record update first, within.. 150sec or so of each other. That solved it but i got a feeling this is somewhat muddy.