Getting error when rejoining a standby to replication cluster
Closed this issue · 1 comments
So I am testing with simple 2 nodes cluster. One being PG-Master and other being PG-replica.
PostgreSQL : 14.2
Ubuntu : 20.04
repmgr : 5.3.1
So I setup replication successfully between two servers. After that, I shut down two servers and start both of them back up again. Now cluster show commands gives below warning.
WARNING: node "PG-replica" not found in "pg_stat_replication"
So I am trying to rejoin PG-replica by running below command
repmgr -f /etc/repmgr/repmgr.conf node rejoin -d 'host=pg-master-ip user=repmgr dbname=repmgr connect_timeout=2'
I am getting below output
NOTICE: rejoin target is node "PG-Master" (ID: 1)
INFO: timelines are same, this server is not ahead
DETAIL: local node lsn is 0/D001310, rejoin target lsn is 0/E0008A0
NOTICE: setting node 2's upstream to node 1
WARNING: unable to ping "host=pg-replica-ip user=repmgr dbname=repmgr connect_timeout=2"
DETAIL: PQping() returned "PQPING_NO_RESPONSE"
NOTICE: starting server using "pg_ctl -w -D '/var/lib/postgresql/14/main' start"
ERROR: unable to start server
ERROR: NODE REJOIN failed
I feel that repmgr is trying to start postgresql using pg_ctl which is not available in debian package installation of postgresql. instead wrapper pg_ctlcluster is used.
Am I doing something wrong here? Please help.
This is not an issue. service_(start|stop|reload|status)_command setting can be used in repmgr.conf to set the custom command for PostgreSQL cluster start/stop. In case of debian, you can use pg_ctlcluster instead of pg_ctl.