Database Reference
In-Depth Information
If replication drops or needs to be restarted, you can use the following command
on the subscriber:
londiste.py clustname_londiste.ini repair table1
See also
http://skytools.projects.postgresql.org/doc/
Thanks to
Marko
Kreen
and
Dimitri
Fontaine
for additional advice, especially
http://wiki.
postgresql.org/wiki/Londiste_Tutorial
, which was the starting point for most
people's first steps with Londiste.
Selective replication using Slony 2.0
Slony is one of the longest running and best known replication projects for PostgreSQL. It
provides advanced features, though as a result is considered complex by many users.
Slony is also known as Slony-I because there were once plans for a Slony-II, though that
doesn't exist (at least so far). Slony-I was first released as 1.0, though there is now a Version
2.0. So when we talk about Slony 2.0, we mean Slony-I Version 2.0.
Getting ready
If you haven't read the recipes on
Replication
Concepts
and
Replication
best
practice
at the
start of this chapter, please go and read them now. Replication is complex.
The first question to look at is whether to use Slony. Slony still has some advantages, even
alongside streaming replication, for specific-use cases. Some of its advantages are as follows:
F
Slony can replicate data between different PostgreSQL major versions.
F
Slony can replicate data between different hardware or operating systems.
F
Slony can record changes into a file, so that those changes can be replicated in bulk
by some mechanism. This can be used to replicate by regular transfer of tapes, and is
also useful when network connection is not always available. This offers the capability
for periodic updates from remote or mobile users.
F
Slony allows you to perform selective replication.
F
Slony also allows a Master to replicate some tables to one Standby and other tables
to a different Standby. We name that
fan-out
. The reverse is also true: a Standby
can receive tables from multiple Masters. We can name that
fan-in
, or maybe
"
roll-up
" capability.
F
Slony allows relay replication, so that the Master (or Origin) sends changes to a
Standby node (a Subscriber), which then sends it onwards (as a Provider) to other
Standbys (also Subscribers).
