Re: [exim] retry configuration

On Mon, Apr 20, 2009 at 4:34 PM, Dean Brooks <dean@???> wrote:
> On Mon, Apr 20, 2009 at 04:04:17PM +0300, Catalin Constantin wrote:
>
>> The problem is caused by the way EXIM is handling retries.
>
> What you are asking for is for Exim to use the entire email address
> as the retry-database index key instead of just the remote domain name.
>
> There is an option called "retry_use_local_part" which can be enabled
> on a special router that will reference the retry database using the
> entire email address.
>
> A basic example would be:
>
>  special_remote_smtp:
>     driver = dnslookup
>     domains = yahoo.com
>     retry_use_local_part
>     transport = remote_smtp
>
> You do *not* want to use the retry_use_local_part option on your
> _default_ router or transport.  If you do, you'll find that hosts
> that really are down will cause every message to be retried, and
> your queue runners will get massively clogged trying to deliver
> those messages.
>
> If you are only moving a very small number of emails a day (less than
> 100 maybe?), then it may not matter.  Just be forewarned.
We send out more than 1000 emails per day to @yahoo.
Sending to Yahoo DOES NOT work with Exim cause of the retry policy.

Even after setting retry_use_local_part, no luck.
It looks like Lena suggested that this configuration is ignored for
remote transports.

Examining the retry db:
# exim_dumpdb /var/spool/exim retry
shows just ~ 45 entries (one for every Yahoo SMTP server IP)

It should have an entry for EACH mail in the queue waiting to be delivered.

Also in the logs, after i clean the spool DB (retry + others), start
the exim daemon, and monitor the logs:
- ~ 30 tries are attempted for delivery @yahoo
- for the rest emails in the queue i get "R=dnslookup T=remote_smtp
defer (-53): retry time not reached for any host".


Any other hints ?

Thanks

--
Catalin Constantin
Dazoot Software