Ok, I've read through all the documentation and threads and still can't find any real info on how to properly setup online recovery and how it connects to the backend Postgres servers and with what user. I've got a simple setup, one pgpool-II (2.1) server and two backend Postgres servers (8.3.3). For testing purposes I've setup the root user with ssh keys so that user can connect to each backend without supplying a password. I run the command: pcp_recovery_node 2000 server_name 9898 user pass 1 and it errors out with the following message:
2008-09-08 10:56:48 LOG: pid 29909: starting recovering node 1 2008-09-08 10:56:48 ERROR: pid 29909: start_recover: could not connect master node. I've enabled debug logging for ssh on the backend postgres nodes and don't see an attempt anywhere to connect. So my questions are these: 1) How is the recovery_user and recovery_password used? If I set the recovery_user to root does that mean root needs to be configured to ssh into all the backend servers with no password or is it used in another way? 2) Do all the Postgres backend servers need to connect to each other via ssh with no password as well? If so what user, whatever the recovery_user is set to? 3) With what user account will the backend scripts be run as (base-backup.sh, pgpool_recovery, pgpool_remote_start) and does it know to use the user I've defined so that it can connect to the other servers without supplying a password? 4) How do the $1, $2, and $3 variables get set in the base-backup.sh script? When you run pcp_recovery_node does that command try to set those variables or is it a manual process? Any info would be appreciated. Thanks again -- Dave
_______________________________________________ Pgpool-general mailing list [email protected] http://pgfoundry.org/mailman/listinfo/pgpool-general
