Robert Mueller wrote:
1) Server A dies
2) Users are sent to Server B via DNS redirects
3) Server A is restored
4) Mailboxes on Server B are now more 'current' than mailboxes on
Server A
In other words, the failover is easy(ish) but the failback has me
scratching my head. I Think two way replication solves that.
Once you've got Server A back, you have to make sure that when you
start it you've disabled the sync_client and all IMAP/POP/LMTP
connections to Server A, otherwise if you have things happening to
both of them at the same time, you'll be in big trouble since there
won't be an "authorative" server anymore.
Before you change back to the original setup, you have to change the
configuration so that Server B has a sync_client and Server A has a
sync_server, and run replication back the other way until all the
mailboxes on Server A are back up to date with the contents of Server
B (and probably for safety do a sync_client -u on each user), then
switch over back to the original setup.
Interesting. This process also implies you want to stop incoming mail
for a period of time to both servers. Sound right?
--
Scott Russell <lnxgeek@xxxxxxxxxx>
IBM Linux Technology Center, System Admin
----
Cyrus Home Page: http://asg.web.cmu.edu/cyrus
Cyrus Wiki/FAQ: http://cyruswiki.andrew.cmu.edu
List Archives/Info: http://asg.web.cmu.edu/cyrus/mailing-list.html
