El jue, 22-01-2004 a las 17:58, Gary Richardson escribi�: > That is why you are tarring the mail every 30 minutes or some other value to > another server. The whole idea is being able to react to your main mail > server going down. If the mail server crashes you should be able to rebuild > it quickly.
I see. But this approach is almost the same as the one with rsync: copying the changes on the mailboxes to another server. But by syncing them to an already installed and configured server, the switchover is immediate (unplug the cable from one server and plug it to the other one). The re-install of the main server is also simpler this way as all you have to do is copy everything from the backup (incl. system, config, etc) back to the primary server. > The whole point is that doing it this way is far cheaper and far more simple > than trying to replicate file systems and queues and so forth.. Cheaper: you need at least two machines in any case, and in my scenario the backup server doesn't need to be a full-fledged server, as it will only go really online for a short time while we recover the primary one. A desktop PC with decent HW and plenty of HD should be able to handle it. Simple: we already discarded keeping the queues in sync, and replicating the file system is as easy as installing the same packages and syncing the mailboxes (of course, the SMTP/IMAP proxying solution would be better _and_ much more complex). -- Vicente Aguilar <[EMAIL PROTECTED]> Departamento de Sistemas Tlf.: 965 98 71 92 Recursos en la Red, S.L.U. http://www.renr.es
