Tony Finch wrote: > On Tue, 26 Aug 2008, Marc Silver wrote:
>> I'm manually trying to flush the queue with 'exim -v -qff'.
>
> Try omitting the force flags, so that you aren't defeating the retry
> logic. Doing so causes Exim to do unnecessary work.
Sure, but that would also result in old (older than 4d) messages being
left in the queue, when they are skipped due to retry logic, wouldn't
it? I may be wrong with that, but I think exim doesn't abandon delivery
without a last try, or does it?
>> Am I incorrect in assuming that if the internet line is already
>> saturated then I must have enough queue runners?
>
> If that's the case then you have fairly serious problems.
Yes, that may explain why the system is having these problems. Bandwidth
may just not be sufficient for the amount of e-mails flowing through.
Tuning exim of course will not help here.
24k messages, 45GB in size result in 1.9MB average size. With the
mentioned 150k mails a day, this would result in a required bandwidth of
28MBit, roughly. The servers are delivering "170GB of mail per 24 hour
period (limited by bandwidth entirely)", this would result in ~17MBit
available bandwidth.
Something is obviously wrong here. :o)
