kingbaseV8R6读写分离集群主备库时间线不同问题解决

主备时间线不同问题

方法一：提升主库时间线

• 描述：主库时间线小于备库时间线，主库有物理备份，在不重做备机的情况下恢复集群

时间线的确定

#node1：(主库)		node2:(备库)
sys_controldata /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data
	##node1:
	Latest checkpoint TimeLineID:       7
	##node2:
	Latest checkpoint TimeLineID:       8

确定集群状态

repmgr cluster show
    """
    1  | node111 | primary | * running            |
    2  | node112 | standby | ! running as primary |
    WARNING: following issues were detected
      - node "node112" (ID: 2) is registered as standby but running as primary
      - node "node112" (ID: 2) is not attached to its upstream node "node111" (ID: 1)
    """

关闭集群

sys_monitor.sh stop

提升主库时间线

##在主机创建备机识别文件
    touch /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data/standby.signal
##启动主机数据库
	sys_ctl -D /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data start
##执行时间线提升命令
    sys_ctl promote -D /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data/ 

检查主备库时间线

sys_controldata /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data

备库执行执行rejoin 操作，将备库重新加入集群

repmgr -h 192.168.174.111 -U esrep -d esrep -p 54321 node rejoin --force-rewind 

确定集群状态

repmgr cluster show

方法二：降低备库时间线

查看时间线

	##此时备库和主库已不在同一个时间线上
	###备库时间线
    sys_controldata /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data |grep TimeLineID
    Latest checkpoint's TimeLineID:       14
    Latest checkpoint's PrevTimeLineID:   14

	##主库时间线
    sys_controldata /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data |grep TimeLineID
    Latest checkpoint's TimeLineID:       13
    Latest checkpoint's PrevTimeLineID:   13

修复备库

#查看切换点
ls sys_wal/*.history

#查看最新时间线的.history文件
    cat 0000000E.history
    "
    执行结果：
    1       0/90000A0       no recovery target specified
    2       0/C0000A0       no recovery target specified
    3       0/1B0000A0      no recovery target specified
    4       0/1E0000A0      no recovery target specified
    5       0/210000A0      no recovery target specified
    6       0/220000A0      no recovery target specified
    7       0/250000A0      no recovery target specified
    9       0/260000A0      no recovery target specified
    11      0/290000A0      no recovery target specified
    12      0/2A0000A0      no recovery target specified
    13      0/2A00EA30      no recovery target specified
    "
	
#从0000000E000000000000002A开始，所有的wal必须存在备库sys_wal目录中。如果已经覆盖了，必须从归档目录拷贝到备库sys_wal目录中，也可以直接将归档文件全部拷贝到sys_wal目录下
    "
    cp /home/kingbase/kbbr_repo/archive/kingbase/12-1/* /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data/sys_wal
    "  					-- /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/archive为归档目录
    
#停掉备库
	sys_ctl -m fast -D /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data stop

#测试修复是否能够成功
	sys_rewind -n -D /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data/ --source-server="hostaddr=192.168.174.111 user=system port=54321 dbname=test"
	
   " 
   执行结果：
   datadir_source = /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data
   sys_rewind: servers diverged at WAL location 0/2A00EA30 on timeline 13
    sys_rewind: rewinding from last common checkpoint at 0/2A00E988 on timeline 13
    sys_rewind: find last common checkpoint start time from 2022-01-19 15:29:33.495709 CST to 2022-01-19 15:29:33.560238 CST, in "0.064529" seconds.
    sys_rewind: rewind start wal location 0/2A00E958 (file 0000000D000000000000002A), end wal location 0/2A01C500 (file 0000000D000000000000002A). time from 2022-01-19 15:29:33.495709 CST to 2022-01-19 15:29:37.178985 CST, in "3.683276" seconds.
    sys_rewind: Done!
    "

#可以修复，直接修复
	sys_rewind -D /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data/ --source-server="hostaddr=192.168.174.111 user=system port=54321 dbname=test"
    
    "
    执行结果：
    datadir_source = /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data
    sys_rewind: servers diverged at WAL location 0/2A00EA30 on timeline 13
    sys_rewind: rewinding from last common checkpoint at 0/2A00E988 on timeline 13
    sys_rewind: find last common checkpoint start time from 2022-01-19 15:32:25.914143 CST to 2022-01-19 15:32:25.946141 CST, in "0.031998" seconds.
    sys_rewind: update the control file: minRecoveryPoint is '0/2A02AD30', minRecoveryPointTLI is '13', and database state is 'in archive recovery'
    sys_rewind: rewind start wal location 0/2A00E958 (file 0000000D000000000000002A), end wal location 0/2A02AD30 (file 0000000D000000000000002A). time from 2022-01-19 15:32:25.914143 CST to 2022-01-19 15:32:29.572945 CST, in "3.658802" seconds.
    sys_rewind: Done!
    "

#检查备库时间线
	sys_controldata /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data |grep TimeLineID
    Latest checkpoint's TimeLineID:       13
    Latest checkpoint's PrevTimeLineID:   13

#修改配置文件kingbase.auto.conf
    vi kingbase.auto.conf
    primary_conninfo = 'user=esrep connect_timeout=10 host=192.168.174.111 port=54321 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3 application_name=node112'
    recovery_target_timeline = 'latest'
    restore_command = 'cp /home/kingbase/kbbr_repo/archive/kingbase/12-1/%f %p'

#删除错误时间线上产生的归档，即走到时间线0000000E上的归档
    mkdir /home/kingbase/kbbr_repo/archive/kingbase/12-1/error_tl_14
    mv 0000000E* error_tl_14

#创建备库标识文件
	touch /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data/standby.signal

启动备库

sys_ctl start -D /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data/
	##在主库上查看流复制同步状态
    select client_addr,sync_state from sys_stat_replication;
      client_addr   | sync_state 
    ----------------+------------
     192.168.174.112 | async

查看集群状态

repmgr cluster show

1  | node111 | primary | * running |           |
2  | node112 | standby |   running | ! node111 |
WARNING: following issues were detected
  - node "node112" (ID: 2) is not attached to its upstream node "node111" (ID: 1)

将备机加入集群

#关闭备库
sys_ctl stop -D /home/kingbase/cluster/R6C5B23CLS/cls/kingbase/data
#加入集群
repmgr -h 192.168.174.111 -U esrep -d esrep -p 54321 node rejoin --force-rewind 

查看集群状态

1  | node111 | primary | * running |          |
2  | node112 | standby |   running | node111  |

注：数据有风险，操作需谨慎！