Re: [exim] RCPT delays and PIPELINING

On 2/1/07 4:09 AM, "Philip Hazel" <ph10@???> wrote:

> Bottom line: if you are using delays with RCPT commands, consider the
> PIPELINING consequences.

I shut off pipelining shortly after we started our greylisting. I saw
problems with a few sending MTAs I thought related to greylisting. Perhaps
they related to delay instead (my colleague also inserted a couple of short
delays, but far from enough in total to trigger the "5 minute problem").

Either way, those problems went away with no pipelinging.

As we're small enough to have limited impact on the 'Net through turning off
pipelining, I haven't investigated further, although it would seem
friendlier to have it on when possible.

>
> A thought: as I wrote this, it occurred to me that maybe there is a way
> that Exim can take note of having obeyed "delay", and if so, force out
> its output buffer after it has written the next response, instead of
> waiting until it has run out of input. If possible, that would solve
> this issue automatically. Any views on this idea?

Sounds good, if it can be implemented safely and easily. (This or the idea
of flushing upon an executed delay.) Doesn't sound good if it entails major
work to be bulletproof.

Meanwhile, I should take a look at our delays after EHLO response time, to
see whether perhaps they aren't doing what the colleague hoped.

I should also look again at our submission side--extra round trips over
dialup--our dialup submissions reach us over the Internet--or even 256K DSL
are unfortunate. Hmmm.

And back to your initial paragraph: getting this discussion into the
archives is a very good thing.

--John