On 2017-03-19, Rical Jasan <ricaljasan@???> wrote: > On 03/17/2017 09:53 AM, Dan Liles wrote:
>> What version of exim are you using?
>
> On this particular host it's 4.88, but from EPEL:
>
> $ rpm -q exim
> exim-4.88-2.el6.x86_64
>
> I was going to build from source because that package was years-stale,
> but they recently started paying attention to it again.
>
> The per-message keys make sense (see Jeremy's reply), and the problem
> eventually went away. I was thinking if it were absolutely necessary to
> not contact those hosts/domains for a given period, one could probably
> manage that with smarter queue runners. In this case, start one every 4
> hours for recipients in the given domain, and make sure any other queue
> runners don't try to process them if shorter interval runners were
> necessary.
I use every thirty minutes, but I'm running two instances of exim one
with more frequent runs, then I use delivery_event and write a redis
record when I see a trigger phrase in the response.
The redis record holds an ip list excluding the ip address of the server
and is used as ignore_target_hosts in a router to side-track the next
delivery attempt of that, and similar messages to the slow instance.
--
This email has not been checked by half-arsed antivirus software
