[Pgpool-general] What is the recommended way to migrate a DB to replicate it with pgpool

[Felix J. Ogris]

Hi,

> Let's say the postgresql admin is the only user active on the source
> backend during the replication. As WAL is already enabled on that host I
> guess a server restart would make sure that the tx-logs are synced, or

yip, the source node has to be shut down for the pg_xlogs written to disk.

> am I wrong here? Would a dump ( made to the destination backend (not
> using pgpool, but directly through postgresql) then be save, or is the
> WAL-method you've described necessary for some other reason?

If nobody modifies the source database meanwhile, then pg_dump / pg_restore
will do the job. The WAL method just minimizes downtime and time needed for
the actual copy, probably.

Felix