Hi,

 

We're attempting to apply Slony1 to our database for replication.
However, we use a table partitioning system that causes new tables to be
created at runtime. Slony1 doesn't directly support runtime schema
changes, I realize, but we think we have a solution: There's a master
'partition descriptor' table, which has a row for each table in the
system. If we were to create triggers on insert/update/delete for this
table on master and slave, we could, on the master, invoke a trigger
that adds the table to the set and configures it, and on the slave
invoke a trigger that creates the table (since it always follows a
defined template), and subscribes to its replication set.

 

Is this practical? And if so, can anyone give some pointers for getting
started?

 

Thanks,

 

Nick Johnson

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://gborg.postgresql.org/pipermail/slony1-general/attachments/20060330/a9ccf001/attachment.html