PG Efm
main | tips | stats | roles | efm_failover |
restartare standby
Prima su standby, poi master, poi witness
systemctl status efm-dbmnp.service
/usr/efm-2.1/bin/efm cluster-status DBMNP
systemctl stop efm-dbmnp.service
pg_ctl stop -D /app/ologw/pgdata/ppas/db -m fast
per ripristinare DBMNP
sul Master
1. decommentare synchronous_standby_names = 'sync_replication'; in postgresql.conf
2. pg_ctl reload
sullo Slave
restartare db102dbm e vedere se si riallinea altrimenti -> OPI ricostruzione nodo standby
Restart EFM
restartare efm-ctl prima su witness, poi nodo 2 poi nodo 1
systemctl start efm-dbmnp.service
Verificare lo stato Master-StandBy
service
in postgresql.conf
hot_standby=on
edb=# select * from pg_stat_replication;
pid | usesysid | usename | application_name | client_addr | client_hostname | client_port | backend_start | backend_xmin | state | sent_location | write_location | flush_location | replay_location | sync_priority | sync_state
-------+----------+------------+------------------+---------------+-----------------+-------------+---------------------------------+--------------+-----------+---------------+----------------+----------------+-----------------+---------------+------------
21869 | 16650 | edbrepuser | sync_replication | 10.247.153.44 | | 50238 | 25-AUG-17 13:32:35.88557 +02:00 | | streaming | F/74063978 | F/74063978 | F/74063978 | F/74063978 | 1 | sync
(1 row)
edb=# show synchronous_commit ;
synchronous_commit
--------------------
on
(1 row)
edb=# show synchronous_standby_names ;
synchronous_standby_names
---------------------------
sync_replication
(1 row)
N.B. lo host è la macchina primaria. Il recovery.conf è presente sullo StandBy. Verifica anche il recovery.done,
$cat recovery.conf
standby_mode = on
trigger_file = '/tmp/trigger_file'
primary_conninfo = 'host=10.247.153.56 port=5444 user=edbrepuser password=hpsareplica application_name=sync_replication'
recovery_target_timeline = 'latest'
page revision: 1, last edited: 21 Sep 2017 13:17