Re: [Exim] messages not timing out

Richard.Hall wrote:

> I've got exactly the same problem for exactly the same domains, though I
> can only claim 8 days (should have expired at 5 days)
>
> My current (incomplete) analysis goes something like this
>
> a) the messages are NOT frozen (sorry, Nico!)

I just hoped for any easy one. ;)

> b) sinamx.sina.com.cn has 16 (count 'em) A records.
> c) exinext is reporting errors on 40 (yes, really) different addresses
> d) one can only speculate about why their DNS is changing so significantly

They have more than these 16 addresses, your nameserver is just caching
them, and is returning the same subset every time. If you do another
query on a "clean" cache, you'll see another subset.

> e) some of the addresses reported by exinext are very old, eg
...
> g) some are 'past final cutoff time' (I never have managed to understand
> what that means!!), eg

Oh, I'm not the only one. :)

> and yet they are still retried ...

You should really run exim_tidydb once per day.

> I can't help feeling that there is some complicated interaction between
> the number of A records, the relative 'age' of the various addresses, the
> different failure modes, and who knows what else, which is conspiring to
> keep these messages 'alive' beyond their sell-by date. But trying to
> understand the algorithm does my poor little head in, so I'll have to
> leave this to the experts.

I think the problem is that exim only sees a subset of all addresses at
a time. This could "confuse" the retry algorithm.

> I can provide the full exinext output if anyone wants it.

Please send it via PM. Some log-entries of the specific messages and a
debuglog of a forced delivery (exim -d -M MSG-ID) could also be useful.

Nico