Re: [exim] Delay times of less than a second for ratelimitin…

Top Page
Delete this message
Reply to this message
Author: Jeremy Harris
Date:  
To: exim-users
Subject: Re: [exim] Delay times of less than a second for ratelimiting?
On 21/12/16 10:55, Mike Brudenell wrote:
> Ah, OK… That's sad. I was hoping to allow an initial burst of messages
> through (tested with a *ratelimit* condition in the acl_smtp_mail) and then
> use *delay* to slow down throughput to around 250 messages per minute.
> Ignoring reception tie of the message itself, that would require a delay of
> around 60 seconds / 150 = 0.24 seconds.
>
> However if the minimum delay that can be waited is only 1 second that
> effectively caps the ongoing throughput at a rather low 60 messages per
> minute — 60 seconds / 1 second delay = 60 messages per minute.


Um, only if you apply that one second to every message?
Typically you would apply it based on the rate, which would
be averaged over some period and wander each side of your
set-point as a delay was inserted or not-inserted.
--
Cheers,
Jeremy