auguszou / vagrant_pg_master_slave

Postgres master-slave backup

test

1. vagrant up

master

1. login master1, vagrant ssh master1

2. login pg, psql postgres -U postgres -h localhost

3. check replication status, select * from pg_stat_replication;

4. create table create table person(id integer, name varchar(64));

5. insert data insert into person values(1, 'alice');

slave

1. login slave1, vagrant ssh master2

2. login pg, psql postgres -U postgres -h localhost

3. query table select * from person;

setup steps

master

1. update postgresql.conf

listen_addresses = '*'
wal_level = hot_standby
fsync = on
max_wal_senders = 8
wal_keep_segments = 256
wal_sender_timeout = 60s
max_connections = 512
synchronous_commit = local
synchronous_standby_names = '*'
archive_mode = on
archive_command = 'cp %p ${PG_DATADIR}/archive/%f'

2. create deplicate role copydb

create role copydb with replication login encrypted password '${PG_DUPLICATE_PASSWORD}';

3. grant access privileges, add this line to pg_hba.conf

host     replication     copydb          192.168.2.0/24          md5

4. create archive directory, and update owner

mkdir -p ${PG_DATADIR}/archive
chmod 700 ${PG_DATADIR}/archive && chown postgres:postgres ${PG_DATADIR}/archive

5. restart postgres server

/etc/init.d/postgresql restart

slave

1. update postgresql.conf

listen_addresses = '*'
max_connections = 1024
wal_level = hot_standby
hot_standby = on
hot_standby_feedback = on
max_standby_streaming_delay = 30s
wal_receiver_status_interval = 1s
synchronous_commit = local
synchronous_standby_names = ''

2. sync master data directory to slave

SELECT pg_start_backup('mybackup_label', true); # executed in master server pg cmd
rsync -ac --exclude=*pg_xlog* --exclude postmaster.pid root@${PG_MASTER_HOST}:${PG_DATADIR} ${PG_DATADIR} # executed in slave server cmd
SELECT pg_stop_backup(); # executed in master server pg cmd

or

pg_basebackup -D ${PG_DATADIR} -h ${PG_MASTER_HOST} -P -U copydb --xlog-method=stream

3. update ${PG_DATADIR}/recovery.conf

standby_mode = on
recovery_target_timeline = 'latest'
primary_conninfo = 'host=${PG_MASTER_HOST} port=${PG_MASTER_POST} user=copydb password=${PG_DUPLICATE_PASSWORD}'
restore_command = 'cp ${PG_DATADIR}/archive/%f %p'
trigger_file = '${PG_DATADIR}/postgresql.trigger.5432'

4. update postgres data directory owner

chown -R postgres:postgres ${PG_DATADIR}

5. restart master postgres server and slave postgres server

/etc/init.d/postgresql restart
ssh root@${PG_MASTER_HOST} /etc/init.d/postgresql restart