Ok. Can you show the pgpool log and PostgreSQL log(both node 0 and 1) when you did:
>> > pcp_recovery_node 20 localhost 9898 postgres <password> 0 -d -- Tatsuo Ishii SRA OSS, Inc. Japan English: http://www.sraoss.co.jp/index_en.php Japanese: http://www.sraoss.co.jp > Actually, yesterday in desperate moments I have done that, I got > cvs version of pgpool. > I think I have same situation. > I have added failback script that creates trigger file on my new primary > node, > but standby is acting like primary.. > > On Fri, Jan 21, 2011 at 8:19 AM, Tatsuo Ishii <[email protected]> wrote: > >> > I'm having problem with pcp_recovery_node. >> > I have 3 computer architecture. One computer is with installed pg_pool, >> and >> > two other are nodes. >> > Pg_pool version installed is 3.0.1, and postgres is 9.0.2 >> > I have streaming replication configured. >> > >> > I'm trying to setup automatic recovery of nodes. >> > I set one server to be master, and other slave (read only connection). >> > When primary fails, slave is being used as a master and this transition >> from >> > one >> > node to another is automatic and everything works great. >> > But, when I execute pcp_recovery_node command: >> > >> > pcp_recovery_node 20 localhost 9898 postgres <password> 0 -d >> > >> > pgpool is trying to make my first node (one that I'm trying to recover) >> > as a master node, and everything went well except crating trigger file. >> >> That is the limitation of pgpool-II 3.0.x. pgpool-II 3.1 will make >> this possible. You can try the to-be-pgpool-II 3.1 source code from >> CVS HEAD or tar ball from >> >> http://pgpool.projects.postgresql.org/contrib_docs/simple_sr_setting/index.html >> -- >> Tatsuo Ishii >> SRA OSS, Inc. Japan >> English: http://www.sraoss.co.jp/index_en.php >> Japanese: http://www.sraoss.co.jp >> >> > So, in log file there is no errors, failover.sh is not executed, and >> > node1 (just recovered node) is acting like it is master, but it is not >> > until I manually create trigger file. >> > >> > Why pgpool is trying to make my just recovered node as master, and >> > not just bring it up and set it to act like slave. Can I configure this? >> > >> > How can I know what action is provided by executing pcp_recovery_node? >> > >> > I hope that I descibed my problem well. _______________________________________________ Pgpool-general mailing list [email protected] http://pgfoundry.org/mailman/listinfo/pgpool-general
