Graeme Fowler wrote:
On Wed, 2012-10-03 at 13:02 +0100, [email protected] wrote:
As part of the process I intend to shutdown SMTP on mailstoreA in order
to force mail to queue on my Exim routing host - this I am happy with.
OK, but...
However changing the mailstore requires changing the routing data in the
underlying database used by Exim so address [email protected] routes to
mailstoreB - again updating the underlying database is something I am
happy with.
Excuse me going *yick* at this idea :)
If it was me doing this, I'd:
1. bring up B alongside A
2. force A to not deliver locally but smarthost to B
3. change global routing data to route to B
That way you have a transparent changeover without having to take
anything down until you're sure all the mail has drained from A and all
other mail is going to B; then you can shutdown A.
Of course... if A is a closed box with no knobs you can turn to achieve
this, the idea won't float.
Graeme
Seconding Graeme's concept (we've done it in a similar way, and more
than once), and yes it probably CAN float even if all that can be
reached is Exim's config file.
Besides abrupt repointing in the routers, consider also slotting in an
'unseen' that duplicates delivery for a period of parallel-cutover time.
Best proof you can have that nothing is being lost is a comparison
between turn-up timestamp and t+(n) that finds both mailstores match.
Bill
--
韓家標
--
## List details at https://lists.exim.org/mailman/listinfo/exim-users
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://wiki.exim.org/
