Re: round-robin mx and nginx proxy
Richard Hector <richard <at> walnut.gen.nz>
2012-06-22 03:43:19 GMT
On 21/06/12 22:30, Darac Marjal wrote:
> On Thu, Jun 21, 2012 at 03:54:28PM +0800, Umarzuki Mochlis wrote:
>> I need general help on a multi-server setup with
>> 2 MTAs (each also a nginx reverse-proxy)
>> 2 mailbox servers (round-robin)
>> when i check nginx.log on both MTA, only the second MTA got https
>> connection (zimbra support also confirmed this) and it seems that this
>> might caused by DNS setup or related network configuration
>> how do I pinpoint as to where the root cause might be?
> I think I would first start by checking the output of repeated calls to
> "dig +short mx yourdomain.example.org". This should vary, with your
> servers swapping positions each time.
> If you only get the same output each time, then it's down to the sending
> MTA to pick an appropriate MX from the list. I assume both your MXs have
> the same priority; in that case the sending SMTP should pick one at
> random and, if that fails, try the other. Note here, that if you're
> using round-robin AND same-priority MXs, both are picking the order at
Wait - is this https or SMTP? MX records will only affect SMTP, and the
question seemed to relate to https. I got the impression that the fact
that the servers also ran MTAs was incidental.
Still, the issue will relate to what IP address comes back from a query,
and how long it's cached by the client (and intervening name servers).