[Pgpool-general] PGpool II recovery with PITR questions

[Holger Latz]

Hi,

I have a problem with the first stage of recovery of PGpool II. The 
documentation on http://pgpool.projects.postgresql.org/ gives the 
following script in the "Online recovery by PITR" part:

#! /bin/sh

(1) psql -c "select pg_start_backup('pgpool-recovery')" postgres
(2) echo "restore_command = 'scp master:/data/archive_log/%f %p'" \
 > /data/recovery.conf
(3) tar -C /data -zcf pgsql.tar.gz pgsql
(4) psql -c 'select pg_stop_backup()' postgres
(5) scp pgsql.tar.gz secondary:/data

For example node 1 is detached an its database is turned off. Now I want 
to perform the online recovery with the pcp_recovery_node command 
(PGpool II is running on the master-node 0).

Some questions:

a) As far as I understand, the relevant data is copied from the 
master-node to the slave-node. But how does PGpool II recognize, that it 
has to copy from (local) master to a specific slave-backend? Does it 
replace master and slave? I don't think so...

b) Where is the in (2) generated recovery.conf used?

c) I don't find the pgsql file or directory of line (3). I only have 
sth. like 000000010000000000000004 and 
000000010000000000000004.0015AD40.backup in the archive directory.


Cheers
Holger