|
Ok I've found the problem, but I'm at a loss what to do about the solution.
My normal method of ensuring email resiliency is to create 2 MX records, pointing at 2 differing domains hosted on differing ISP's
e.g.
The domain is tech-dept.com
The Primary MX record is balanced.dowmanc1qb.co.uk.
The Secondary MX record is balanced.dowbyf1qb.co.uk.
Dowmanc is hosted in manchester, dowbyf in london.
This has worked great for about 4 years now, with no issues until this past couple of weeks. The reason it turns out, is because MS have implemented a 5 second DNS lookup query in Exchange 2007. If the DNS lookup takes longer than 5 seconds then Exchange 2007 simply puts the email back in the queue and tries later.
If the DNS server hosting dowmanc1qb.co.uk goes down, the query takes longer than 5 seconds. If just the email server based at manchester goes down then the query is pretty much instant, and the failover occurs.
Any suggestions? Every other mail server has the ability to configure the DNS timeout (and most of them are set by default around 20 - 30 seconds).
Should I simply forget trying to implement resiliency to that degree now that MS have decided that 5 seconds is more than enough to do a DNS lookup.
|
Print Thread
deleted
