Hi,
Anyone have thoughts on how I might handle replication with 1 master and 2
slave databases (minimum)?
I am switching our political database from mysql to postgres and need a
solid way to replicate data from our internal server to our webserver. We
currently use 2 databases per server, one administrative database, one live
database for the web. A perl script approximates replication by applying
changes in the local administrative database to the local live database as
well as the administrative databases on the other servers. The catch is
only the changes marked as release changes are sent to the live database. I
suppose I really shouldn't call it replication!
Now that I am doing a extensive redesign of the database and the data entry
client, I am considering consolidating the live and administrative
databases into one with a 'released' field per table to indicate what's
live and what's internal, but this sounds like it would really slow down
the queries. I could use a trigger to dump all the changes into another
table, and then have a perl script to send these changes to the
slaves...and a cronjob on the slaves that removes the records marked as
internal. Ideas/caveats?
Thanks for any help!
Clinton