RFC about 9.3.4 - first query to >=bind.3 caching server always gives sorted order for round robin set
pk at DENIC.DE
Thu Feb 8 21:12:45 UTC 2007
On Thu, Feb 08, 2007 at 10:45:53AM +0100, Stefan Schmidt wrote:
> There you lost me. Why couldn't the MTA just rotate the set itself?
> We're not speaking IN MX here, just plain IN A.
exactly, it's a matter of layering. The resolver is responsible for applying
any topological knowledge it may or may not have to the A RRSet and hand
it off to the application. The MTA can deal with its part by choosing
the MX RR. RFC 2821 (section 5, top three paragraphs on page 61)
specifies that multiple MX RRs with the same preference have to be
processed in random order (which is what you're looking for).
Admittedly, I don't know how well this part is implemented.
More information about the bind-workers