Re: [Exim] No locking problem for hints databases

On Wed, Oct 15, 2003 at 03:31:36PM +0200, Michael Haardt wrote:
> Hello,
> a short while ago I wondered how much avoiding hints databases would
> gain to speed up local deliveries, where hints are stored on tmpfs.
> I measured how long Exim has to wait for the DB lock and found out,
> that even with about 20 queue runners working on a queue of about 5000
> mails, the lock could mostly be taken without measurable delay, a very
> few times the delay was 10ms and just once it was higher: 140 ms.
>
> So the answer for my systems is: Avoiding the hints DB would not help
> at all.

Hi Michael,

I have a few question if you don't mind.
Of the 5000 messages, how many were delivered first time? Specifically,
we were talking about situations where a very small percentage (less
than 15%? Maybe less than 5%) could be delivered first time.
At the same time, were there new messages coming in to the queue?
A good test would need to have constantly new messages coming in.

Please let me know, I'm also in the process of setting up a test
environment for this!