本文首发于 2016-04-25 20:59:52
在使用配置了热备的 PostgreSQL 数据库时,在执行大量事务时,尤其是一个需要插入几千万条数据的 insert 事务时(典型的做法是持续 insert into t select * from t;
),后台 csv log 中报错如下:
2015-07-01 13:25:29.430 CST,,,27738,,51d112c8.6c5a,1,,2015-07-01 13:25:28 CST,,0,LOG,00000,"streaming replication successfully connected to primary",,,,,,,,"libpqrcv_connect, libpqwalreceiver.c:171",""
2015-07-01 13:25:29.430 CST,,,27738,,51d112c8.6c5a,2,,2015-07-01 13:25:28 CST,,0,FATAL,XX000,"could not receive data from WAL stream:FATAL: requested WAL segment 0000000800002A0000000000 has already been removed
",,,,,,,,"libpqrcv_receive, libpqwalreceiver.c:389",""
根据报错信息分析,推测是主库大事务产生了大量 xlog,这是因为 PostgreSQL 在执行事务过程中,直到提交时才会发送到备库。
由于该事务需要执行的时间过长,超过了 checkpoint 的默认间隔,所以导致有的 xlog 还未发送到备库却被 remove 掉了。
要解决该问题,一般可用的方案有:
将 GUC 参数 wal_keep_segments
设大一些,比如设置为2000,而每个 segment 默认值为16MB,就相当于有 32000MB,那么,最多可保存 30GB 的 xlog ,超过则删除最早的 xlog 。
不过,该方法并不能从根本上解决该问题。毕竟,在生产环境中或TPCC等测试灌数时,如果某条事务需要插入几十亿条记录,有可能还是会出现该问题。
归档,就是将未发送到备库的 xlog 备份到某个目录下,待重启数据库时再将其恢复到备库中去。
GUC 参数设置示例如下:
wal_level = hot_standby
archive_mode = on
archive_command = 'rsync -zaq %p postgres@pg-slave:/var/lib/pgsql/wal_restore/%f && test ! -f /var/lib/pgsql/backup/wal_archive/%f && cp %p /var/lib/pgsql/backup/wal_archive/'
archive_timeout = 300
max_wal_senders = 5
wal_keep_segments = 0
wal_level = hot_standby
archive_mode = on
archive_command = 'test ! -f /var/lib/pgsql/backup/wal_archive/%f && cp -i %p /var/lib/pgsql/backup/wal_archive/%f < /dev/null'
hot_standby = on
wal_keep_segments = 1
standby_mode = 'on'
primary_conninfo = 'host=pg-master port=5432 user=replicator'
restore_command = 'cp /var/lib/psql/wal_restore/%f %p'
archive_cleanup_command = 'pg_archivecleanup /var/lib/pgsql/wal_restore/ %r'
该方法是根本解决方法,不会造成xlog的丢失。也就是说,在 xlog 被拷贝到从库之前,主库不会删除。
启用方法:
max_replication_slots = 2000
postgres=# SELECT * FROM pg_create_physical_replication_slot('node_a_slot');
slot_name | xlog_position
-------------+---------------
node_a_slot |
postgres=# SELECT * FROM pg_replication_slots;
slot_name | slot_type | datoid | database | active | xmin | restart_lsn
-------------+-----------+--------+----------+--------+------+-------------
node_a_slot | physical | | | f | |
(1 row)
standby_mode = 'on'
primary_conninfo = 'host=192.168.4.225 port=19000 user=wslu password=xxxx'
primary_slot_name = 'node_a_slot'
https://www.postgresql.org/docs/9.4/static/runtime-config-replication.html
https://www.postgresql.org/docs/9.4/static/warm-standby.html#CASCADING-REPLICATION
http://blog.2ndquadrant.com/postgresql-9-4-slots/
http://grokbase.com/t/postgresql/pgsql-general/13654jchy3/trouble-with-replication
http://stackoverflow.com/questions/28201475/how-do-i-fix-a-postgresql-9-3-slave-that-cannot-keep-up-with-the-master
欢迎关注我的微信公众号【数据库内核】:分享主流开源数据库和存储引擎相关技术。
标题 | 网址 |
---|---|
GitHub | https://dbkernel.github.io |
知乎 | https://www.zhihu.com/people/… |
思否(SegmentFault) | https://segmentfault.com/u/db… |
掘金 | https://juejin.im/user/5e9d3e… |
开源中国(oschina) | https://my.oschina.net/dbkernel |
博客园(cnblogs) | https://www.cnblogs.com/dbkernel |