On Mon, 10 Jan 2000 roger@??? wrote:
> 4. Most times exim starts delivering to all recipients on the list within
> one minute of the above. Sometimes however, the message sits in the mail
> queue for silly lengths of time, and then suddenly without anything being
> changed, it starts sending! The silly length of them can be anywhere
> between 10 mins and an hour. During the time it sits there, other email
> for the same list comes in and gets sent off immediately. Forcing a queue
> run doesn't help, as exim just reports that the spool file is locked.
>
> What I have noticed is that when this happens, the contents of the
> ../spool/input directory contain 3 files for this message, -D, -H and -J.
> I know the D & H are the data and header files, but I'm not sure what the
> J file is.
The J file is a delivery journal, which gets created at the start of a
delivery process (and removed at the end). This suggests that a delivery
process has started, but is being held up for some reason. How long is
the majordomo list? Exim does all the routing before it does any actual
deliveries. Consequently, if it's a long list and there are DNS delays,
it could hold it up a while, but that doesn't make sense if other
messages to the same list go through.
> I assume that when majordomo delivers the email to exim , a process is
> started for it (correctly) but something is making the process just sit
> there doing nothing...
Yes. One way of finding out might be to attach a debugger (strace,
truss, whatever you have) to the process to see where it is sitting.
> If copies of my logs when this happens will help, I'll send some.
Probably not very helpful. What is needed is -d9 output from a failing
delivery, but I can see that this isn't something that is easy to
reproduce...
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
