Monday, November 24, 2003, 5:30:05 PM, Erik wrote:

EB> You're over-engineering.  DRDB and Intermezzo are not ready for this.  Try to
EB> imagine recovering from a failure in either of these.
ok, intermezzo really seems to be not ready for production, but drbd
makes a pretty stable impression on me. i have set up two boxes and
will test the system tomorrow.

EB> I suggest simply having every mail deliver to both locations,
how can i accomplish this? the only way would be a some kind of
balancer in front of the the qmail boxes, but then again i have
another spof.
EB> and hacking a pop/imap server that deletes from both.
ARG... i guess this exceeds my abilities :)

EB> Personally, I don't go that far; I
EB> just deliver to both and have rsync clean up the second member of the storage
EB> pair when a delete happens.
how do you do this double-delivery?

EB> Consider separating your MXs away from your mail storage boxes.
EB> Really, though, all you need is one stable box.  Don't buy scummy IDE drives.
EB> You'll have a better time with one really good machine than with two crashy
EB> ones, and spend less.
im just about to migrate this 500 accs from one crashy box to two of
them :) no, i think you just need some sort of redundancy to achieve
a certain level of availability. and i thought, this would be a rather "cheap"

thanks for your opinion


Reply via email to