Forum Moderators: phranque
I'm working for a company ("A"). Our emails ("myolddomain") are dealt with by provider ("P"), but we had a domain ("mynewdomain") (administered by "E") with forwarding to myolddomain.
This far from adequate situation worked ok for a few years. The problem came just over a week and a half ago when some emails from a Singaporean domain (A's clients) began to bounce when sent to mynewdomain. A little investigation showed that emails sent directly to myolddomain were received fine, as were emails sent to mynewdomain from, for example, one of the popular free webmail accounts. The problem appears to only affect certain companies, all in Singapore.
Now, a little more investigation revealed that one of E's servers was rejecting the client's emails, with a "550 relay not allowed" error - apparently the domain is greylisted.
Using this as an opportunity to do what should have been done a long time ago, I switched hosting of the email to another provider ("S"), changed the MX records, and sat back and waited. And it was fine: all diagnostic tools showed my test emails to be going to the new servers. Not so the Client. His emails still bounced.
I took another step (not sure this would've made any difference, but I gave it a go), and changed to using S's name servers for all functions. Now I assumed that E's naughty servers would be completely out of the loop. Fine - worked with all my diagnostic tools. Didn't work for the client's emails, which were still going to E's servers and, predictably, getting rejected.
A little more diagnostic work took place, and some kind of trace was done at the client's end. A section of the returned transcript read something like this:
06302006 12:08:18:Processing mynewdomain.com
06302006 12:08:18:Retrieving records from local cache for MX
Data:<mynewdomain.com>.
06302006 12:08:18:Lookup Returned <[(10, 'mailserver1.providerE.co.uk',
>> ('#*$!.#*$!.#*$!.xxx',)), (10, 'mailserver2.providerE.co.uk', ('xxx.xxx.xxx.xxx',)),
(10,
'mailserver3.providerE.co.uk'', ('xxx.xxx.xxx.xxx',)), (10, 'mailserver4.providerE.co.uk'',
('xxx.xxx.xxx.xxx',)), (10, ''mailserver5.providerE.co.uk'', ('xxx.xxx.xxx.xxx',))]>
So, the pertinent words in the above transcript appear, to my untrained eye, to be "local" and "cache". It is my (untrained...) opinion that flushing the client's local MX cache would likely solve the problem.
Now, please, is this about right? What else could have gone wrong? Have I set something else up incorrectly (though I know of no other problems from other sources)? Help!
Any comment/advice would be welcome.
Good times.
