[Bucardo-general] Newbie Questions

[David E. Wheeler]

Hello Bucardoans,

I'm looking at cross-data center multimaster replication solutions for $work, and so am naturally trying to understand Bucardo. I've installed the latest 4.99.x code from Git and worked through Greg's brief tutorial here:

  http://blog.endpoint.com/2011/06/bucardo-multi-master-for-postgresql.html

However, the documentation is a bit thin. The man page is great, but assumes that one already understands things pretty well. So I've collected a bunch of questions below. Once I have a couple days of this stuff under my belt and understand the fundamentals, I'll blog it and/or document it in the Wiki.

Anyway, to the questions!

Bucardo Questions

• What are these programs and what are their responsibilities?
	• Master Control Program (MCP)
	• Kids
	• Controller
	• VAC (Not mentioned by `stop` command docs in the man page)
• How many connections will Bucardo use?
• Can Bucardo connect to a connection pooler, e.g., pgBouncer?
• Is there a glossary for these terms?
	• dbs
	• dbgroups
	• tables
	• sequences
	• syncs
	• herds
	• customnames
	• customcols
	• customcode
	• ping
• Are there Linux start scripts to keep this thing running in production?
• Has anyone designed a cross-data center multi master replication configuration?
• What should such a thing look like?
• Can I have multiple Bucardos running, or is it a single point of failure?
• Is there some sort of failover support in case Bucardo dies?

Where this is headed: We have two data centers, and want to be able to shut one down (or have it go down) without interfering with the other at all. I imagine that if only one Bucardo can be running to replicate between masters in each data center, then if the box with that Bucardo falls over, we need some way to have another come up ASAP. I figure it's less of an issue if we take stuff down for maintenance, as it will just sync when it comes back up. AMIRITE?

Thanks,

David