On 02/11/2016 12:03 AM, Graeme Fowler wrote:
> On 9 Feb 2016, at 14:29, Konstantin Boyandin <lists@???> wrote:
>> The above Exim behaviour has happened just few times. Nameservers'
>> responses are being monitored 24/7 and they didn't show any quirks.
>
> With the same granularity that Exim has? Are the resolvers the same in
each case?
Every 5 minutes. Affirmative, same resolvers.
If I make it significantly more frequent, VM will spend most if its time
wasting bandwidth and CPu time on checking DNS records
>> Question: does Exim uses /etc/hosts records (as specified in
>> /etc/nsswitch.conf ) before trying DNS resolution? Looks like it
doesn’t.
>
> Not for an MX record, no - there’s no way to indicate an MX in
/etc/hosts.
Actually, after I understood it tries to communicate with A record, I
attempted to redirect the A record to correct IP (since domain actual A
record isn't ever used from that VM).
> It *might* drop back to that for A lookups, depending on the lookup order
in /etc/nsswitch.conf.
Corresponding entry:
hosts: files dns
Can the above Exim behaviour be changed? If MX entries fail to resolve,
I would rather expect Exim delay the delivery until DNS returns to
normal functioning. Sending to A is anyway senseless in many cases. Or,
as alternative, add a parameter - host name to fallback if DNS
resolution fails.
Thanks.
Sincerely,
Konstantin
