FATAL: requested WAL segment has already been removed:PostgreSQL 流复制、同步复制卡住和复制槽实测

  pg_basebackup 退出码 0 ;standby.signal 存在: True ;postgresql.auto.conf 里的 primary_conninfo: host=127.0.0.1 port=9261(conninfo 里的其余字段略)
  备库启动退出码 0
  备库 pg_is_in_recovery() = True
  主库 pg_stat_replication:state = streaming ,sync_state = async
  在备库上建表: ReadOnlySqlTransaction: cannot execute CREATE TABLE in a read-only transaction

  30 次单行提交:备库可见延迟 中位数 0.15 ms,最大 1.10 ms
  主库一次性插入 50 万行后:备库回放落后的字节数最大 31800 KB,约 0.06 s 追平;pg_stat_replication 最后一次的 write_lag / flush_lag / replay_lag = 5.6 ms / 5.6 ms / 32.3 ms

  异步复制:单行提交延迟中位数 0.106 ms
  设了 synchronous_standby_names = '*' 之后 sync_state = sync
  同步复制(remote_write 之上的默认 on):单行提交延迟中位数 0.186 ms
  remote_apply(等备库回放完成):单行提交延迟中位数 0.130 ms
  备库停掉 2 秒后,主库上这条 insert 的提交仍然没有返回: is alive
  pg_stat_activity 里它的状态: ('active', 'IPC', 'SyncRep')
  给它发 cancel 之后:2.0 s 时返回 → returned
      WARNING: canceling wait for synchronous replication due to user request | DETAIL: The transaction has already committed locally, but might not have been replicated to the standby.
  这行数据在主库上已经提交了: 1 行(但没有确认到达备库)

  备库停机期间主库又写了 264 MB WAL,做了 8 次检查点,主库 pg_wal 里现在只剩 4 个段文件(min_wal_size = 32MB,max_wal_size = 64MB)
  再启动备库,退出码 0
      LOG:  started streaming WAL from primary at 0/<lsn> on timeline 1
      FATAL:  could not receive data from WAL stream: ERROR:  requested WAL segment <segment> has already been removed
      LOG:  started streaming WAL from primary at 0/<lsn> on timeline 1
      FATAL:  could not receive data from WAL stream: ERROR:  requested WAL segment <segment> has already been removed

  备库停机、主库又写入 60 万行并做了 6 次检查点之后:slot s1,active = False,wal_status = extended,为它保留的 WAL = 198 MB
  主库 pg_wal 里现在有 13 个段文件(不设复制槽时只有 4 个)
  再启动备库,退出码 0
  备库追平后 t 的行数: 1900631 ,主库: 1900631
  追平后 slot s1:active = True,保留的 WAL = 0 bytes