I've thought of this, but don't think this would work...
It would work for the first day, but break the second day, so you would need a 'clean' box everyday. I have one setup that takes the CMU backup and transfers it to another standby server, but does not import. In the event of a disaster, you would just have to run the import at that time... The RTO (return to operation) is higher, like an hour or so for the import process to complete or maybe less/more depending on number of vsites/users, than if it was a true fail-over setup, but was acceptable for this particular client.
The cmuImport would have to "merge" and "default to yes" as options, to add new users to existing sites and replace mail with the new mail. I don't see any options to set cmuImport to do this by default... Also, I don't think there is any function currently built in that would remove users with cmuImport, nor to remove files or an option to completely replace website files and I cannot remember how a merge works with the website files.
There was an older discussion on this list that is looking at replicating data directly between cced's and then maybe rsync with the files? While it was about clustering, you may still want to investigate this if you need HA and/or lower RTO. I know it was just in the beginning discussion phase, but since it was back in Nov '05, there may have been progress since...
The thread was "Possible Roadmap for BlueQuartz?"
http://bluequartz.org/ml/archive/coba-e/3400/3404.html
If there has been any progress on this?
_____________________
Rusty Waybrant
rwaybrant (at mark) gramtel.net
-----Original Message-----
From: Colin Jack [mailto:colin (at mark) mainline.co.uk]
Sent: Monday, June 19, 2006 10:55 AM
To: coba-e (at mark) bluequartz.org
Subject: [coba-e:05797] Redundancy
I am not an linux expert, but was wondering whether in principal the following is 'do-able' ...
Have CMU backup to an FTP server every night and then have another backup server running cmuImport every night on a cronjob to duplicate the other server. Then if the primary goes down we could have some sort of heartbeat monitor that switches across to the backup. Yes, users may get some dupe mail, but apart from that everything should carry on before ... or am I being na$B}W(Be here?
Colin