KingbaseES R6 集群 recovery 参数对切换的影响
案例说明：在KingbaseES R6集群中，主库节点出现宕机（如重启或关机），会产生主备切换，但是当主库节点系统恢复正常后，如何对原主库节点进行处理，保证集群数据的一致性和安全，可以通过对repmgr.conf文件中配置recovery参数来解决。
本案例记录了‘recovery’参数的三种配置情况下，primary 主机重启后，集群恢复的过程。
注意：对于KingbaseES R6老的版本，recovery参数只支持‘manual’和‘automatic’。
数据库版本：
集群架构：
集群节点信息：
案例一：测试‘recovery = standby’
一、执行主备切换测试
1、配置recovery参数（所有node）：
2、查看集群节点状态信息
[kingbase@node1 bin]$ ./repmgr cluster show
ID | Name | Role | Status | Upstream | Location | Priority | Timeline | Connection string
----+---------+---------+-----------+----------+----------+----------+----------+---------------------------
1 | node243 | primary | * running | | default | 100 | 3 | host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
2 | node248 | standby | running | node243 | default | 100 | 3 | host=192.168.7.248 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
3、主库节点系统重启
[root@node3 ~]# reboot
4、查看备库hamgr日志
=从hamgr日志获知，原主库宕机后，集群主备切换，原备库提升为主库。=
[kingbase@node1 log]$ tail -f 100 hamgr.log
tail: cannot open ‘100’ for reading: No such file or directory
==> hamgr.log <==
[2022-03-01 13:12:23] [NOTICE] repmgrd (repmgrd 5.0.0) starting up
[2022-03-01 13:12:23] [INFO] connecting to database "host=192.168.7.248 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3"
INFO: set_repmgrd_pid(): provided pidfile is /home/kingbase/cluster/R6C5/R6C5R/kingbase/etc/hamgrd.pid
[2022-03-01 13:12:23] [NOTICE] starting monitoring of node "node248" (ID: 2)
[2022-03-01 13:12:23] [INFO] "connection_check_type" set to "ping"
[2022-03-01 13:12:23] [INFO] monitoring connection to upstream node "node243" (ID: 1)
[2022-03-01 13:12:23] [NOTICE] try to change wal catched_up state to 1
[2022-03-01 13:12:23] [INFO] primary flush lsn is 0/12000900, local flush lsn is 0/12000848
[2022-03-01 13:12:23] [NOTICE] try to change streaming_sync state to TRUE
[2022-03-01 13:17:24] [INFO] node "node248" (ID: 2) monitoring upstream node "node243" (ID: 1) in normal state
[2022-03-01 13:20:00] [WARNING] unable to ping "host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3"
[2022-03-01 13:20:00] [DETAIL] PQping() returned "PQPING_REJECT"
[2022-03-01 13:20:00] [WARNING] unable to connect to upstream node "node243" (ID: 1)
[2022-03-01 13:20:00] [INFO] sleeping 6 seconds until next reconnection attempt
[2022-03-01 13:20:06] [INFO] checking state of node 1, 1 of 10 attempts
[2022-03-01 13:20:16] [WARNING] unable to ping "user=esrep connect_timeout=10 dbname=esrep host=192.168.7.243 port=54321 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3 fallback_application_name=repmgr"
[2022-03-01 13:20:16] [DETAIL] PQping() returned "PQPING_NO_RESPONSE"
[2022-03-01 13:20:16] [INFO] sleeping 6 seconds until next reconnection attempt
......
[2022-03-01 13:21:23] [INFO] checking state of node 1, 10 of 10 attempts
[2022-03-01 13:21:23] [WARNING] unable to ping "user=esrep connect_timeout=10 dbname=esrep host=192.168.7.243 port=54321 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3 fallback_application_name=repmgr"
[2022-03-01 13:21:23] [DETAIL] PQping() returned "PQPING_NO_RESPONSE"
[2022-03-01 13:21:23] [WARNING] unable to reconnect to node 1 after 10 attempts
[2022-03-01 13:21:23] [NOTICE] setting "wal_retrieve_retry_interval" to 86405000 milliseconds
[2022-03-01 13:21:23] [WARNING] wal receiver not running
[2022-03-01 13:21:23] [NOTICE] WAL receiver disconnected on all sibling nodes
[2022-03-01 13:21:23] [INFO] WAL receiver disconnected on all 0 sibling nodes
[2022-03-01 13:21:23] [INFO] 0 active sibling nodes registered
[2022-03-01 13:21:23] [INFO] primary and this node have the same location ("default")
[2022-03-01 13:21:23] [INFO] no other sibling nodes - we win by default
[2022-03-01 13:21:23] [NOTICE] setting "wal_retrieve_retry_interval" to 5000 ms
[2022-03-01 13:21:23] [NOTICE] this node is the only available candidate and will now promote itself
[2022-03-01 13:21:23] [INFO] try to ping the trusted_servers "192.168.7.1" before execute promote_command
[2022-03-01 13:21:25] [NOTICE] PING 192.168.7.1 (192.168.7.1) 56(84) bytes of data.
--- 192.168.7.1 ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1000ms
rtt min/avg/max/mdev = 2.324/6.238/10.152/3.914 ms
[2022-03-01 13:21:25] [NOTICE] successfully ping one or more of the trusted_servers "192.168.7.1"
[2022-03-01 13:21:26] [NOTICE] try to stop old primary db (host: "192.168.7.243")
[2022-03-01 13:21:26] [NOTICE] PING 192.168.7.241 (192.168.7.241) 56(84) bytes of data.
--- 192.168.7.241 ping statistics ---
2 packets transmitted, 0 received, +1 errors, 100% packet loss, time 1000ms
[2022-03-01 13:21:26] [WARNING] ping host"192.168.7.241" failed
[2022-03-01 13:21:26] [DETAIL] average RTT value is not greater than zero
[2022-03-01 13:21:26] [INFO] loadvip result: 1, arping result: 1
[2022-03-01 13:21:26] [NOTICE] new primary node (ID: 2) acquire the virtual ip 192.168.7.241/24 success
[2022-03-01 13:21:26] [INFO] promote_command is:
"/home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/repmgr standby promote -f /home/kingbase/cluster/R6C5/R6C5R/kingbase/etc/repmgr.conf"
NOTICE: promoting standby to primary
DETAIL: promoting server "node248" (ID: 2) using sys_promote()
NOTICE: waiting up to 60 seconds (parameter "promote_check_timeout") for promotion to complete
NOTICE: STANDBY PROMOTE successful
DETAIL: server "node248" (ID: 2) was successfully promoted to primary
[2022-03-01 13:21:30] [INFO] switching to primary monitoring mode
[2022-03-01 13:21:30] [NOTICE] monitoring cluster primary "node248" (ID: 2)
[2022-03-01 13:21:30] [INFO] create a thread 0x7fe7dbe15700 to check the cluster status
[2022-03-01 13:21:30] [INFO] node (ID: 1): no server running
[2022-03-01 13:21:31] [INFO] [thread 0x7fe7dbe15700] the cluster has no other running primary node, exit
二、原主库节点系统恢复后加入集群测试
1、在新主库创建replication slot
test=# select sys_create_physical_replication_slot('repmgr_slot_1');
sys_create_physical_replication_slot
--------------------------------------
(repmgr_slot_1,)
(1 row)
test=# select sys_create_physical_replication_slot('repmgr_slot_2');
sys_create_physical_replication_slot
--------------------------------------
(repmgr_slot_2,)
(1 row)
test=# select * from sys_replication_slots;
slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn
---------------+--------+-----------+--------+----------+-----------+--------+------------+------+--------------+-------------+---------------------
repmgr_slot_1 | | physical | | | f | f | | | | |
repmgr_slot_2 | | physical | | | f | f | | | | |
(2 rows)
2、原主库系统启动完成：
1）备份新备库节点数据目录
[kingbase@node3 kingbase]$ cp -r data data.bk
2）在data下创建备库标识文件（重要）
[kingbase@node3 data]$ touch standby.signal
3）查看新备库连接字串信息
[kingbase@node3 data]$ cat kingbase.auto.conf
# Do not edit this file manually!
# It will be overwritten by the ALTER SYSTEM command.
job_queue_processes = '5'
primary_conninfo = 'user=esrep connect_timeout=10 host=192.168.7.248 port=54321 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3 application_name=node243'
recovery_target_timeline = 'latest'
primary_slot_name = 'repmgr_slot_1'
wal_retrieve_retry_interval = '5000'
synchronous_standby_names = '1 (*)'
wal_retrieve_retry_interval = '5000'
4）启动新备库数据库服务
kingbase@node3 bin]$ ./sys_ctl start -D ../data
......
NOTICE: standby node "node243" (ID: 1) successfully registered
5）查看当前集群节点状态
[kingbase@node3 bin]$ ./repmgr cluster show
ID | Name | Role | Status | Upstream | Location | Priority | Timeline | Connection string
----+---------+---------+----------------------+----------+----------+----------+----------+----------------
1 | node243 | primary | ! running as standby | | default | 100 | 3 | host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
2 | node248 | standby | ! running as primary | | default | 100 | 4 | host=192.168.7.248 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
WARNING: following issues were detected
- node "node243" (ID: 1) is registered as primary but running as standby
- node "node248" (ID: 2) is registered as standby but running as primary
6）集群自动恢复新备库
=如下hamgr日志所示，启动新备库数据库服务后，集群自动对备库做recovery，并将原主库以备库的模式加入集群。=
*[2022-03-01 13:26:31] [INFO] monitoring primary node "node248" (ID: 2) in normal state
[2022-03-01 13:27:28] [INFO] child node: 1; attached: no
[2022-03-01 13:27:28] [INFO] check node status again, try 1 / 10 times
[2022-03-01 13:27:30] [INFO] child node: 1; attached: no
.....
[2022-03-01 13:27:46] [INFO] check node status again, try 10 / 10 times
[2022-03-01 13:27:48] [INFO] child node: 1; attached: no
[2022-03-01 13:27:48] [INFO] found node down, recovery will be triggered after recovery delay time 20s
[2022-03-01 13:27:50] [INFO] child node: 1; attached: no
......
[2022-03-01 13:28:08] [INFO] child node: 1; attached: no
[2022-03-01 13:28:08] [INFO] recovery delay time reached. can do recovery now.
[2022-03-01 13:28:09] [NOTICE] mark node "node243" (ID: 1) as inactive
[2022-03-01 13:28:09] [INFO] [thread pid:30763] do_nodes_recovery thread begin. The pthread_t tid is 0x7fe7dbe15700
[2022-03-01 13:28:09] [NOTICE] [thread pid:30763] node (ID: 1; host: "192.168.7.243") is not attached, ready to auto-recovery
[2022-03-01 13:28:09] [NOTICE] [thread pid:30763] Now, the primary host ip: 192.168.7.248
[2022-03-01 13:28:10] [INFO] [thread pid:30763] ES connection to host "192.168.7.243" succeeded, ready to do auto-recovery
[2022-03-01 13:28:10] [NOTICE] kbha: node (ID: 1) is running as standby, stop it and do rejoin.
[2022-03-01 13:28:15] [INFO] unlink file /tmp/.s.KINGBASE.54321.lock
[2022-03-01 13:28:15] [NOTICE] executing repmgr command "/home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/repmgr --dbname="host=192.168.7.248 dbname=esrep user=esrep port=54321" node rejoin --force-rewind"
NOTICE: sys_rewind execution required for this node to attach to rejoin target node 2
DETAIL: rejoin target server's timeline 4 forked off current database system timeline 3 before current recovery point 0/130000A0
NOTICE: executing sys_rewind
DETAIL: sys_rewind command is "/home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/sys_rewind -D '/home/kingbase/cluster/R6C5/R6C5R/kingbase/data' --source-server='host=192.168.7.248 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3'"
sys_rewind: servers diverged at WAL location 0/12000A08 on timeline 3
sys_rewind: rewinding from last common checkpoint at 0/11000058 on timeline 3
sys_rewind: find last common checkpoint start time from 2022-03-01 13:28:15.600702 CST to 2022-03-01 13:28:16.200048 CST, in "0.599346" seconds.
sys_rewind: update the control file: minRecoveryPoint is '0/12011F70', minRecoveryPointTLI is '4', and database state is 'in archive recovery'
*sys_rewind: rewind start wal location 0/11000028 (file 000000030000000000000011), end wal location 0/12011F70 (file 000000040000000000000012). time from 2022-03-01 13:28:15.600702 CST to 2022-03-01 13:28:36.045129 CST, in "20.444427" seconds.
sys_rewind: Done!
NOTICE: 0 files copied to /home/kingbase/cluster/R6C5/R6C5R/kingbase/data
NOTICE: setting node 1's upstream to node 2
WARNING: unable to ping "host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3"
DETAIL: PQping() returned "PQPING_NO_RESPONSE"
NOTICE: begin to start server at 2022-03-01 13:28:36.437003
NOTICE: starting server using "/home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/sys_ctl -w -t 90 -D '/home/kingbase/cluster/R6C5/R6C5R/kingbase/data' -l /home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/logfile start"
NOTICE: start server finish at 2022-03-01 13:28:37.367954
NOTICE: NODE REJOIN successful
DETAIL: node 1 is now attached to node 2
[2022-03-01 13:28:38] [NOTICE] kbha: node (ID: 1) rejoin success.
[2022-03-01 13:28:38] [NOTICE] [thread pid:30763] node "node243" (ID: 1) auto-recovery success
[2022-03-01 13:28:38] [INFO] [thread pid:30763] do_nodes_recovery thread ends. The pthread_t tid is 0x7fe7dbe15700
[2022-03-01 13:28:39] [INFO] SET synchronous TO "sync" on primary host
[2022-03-01 13:28:39] [INFO] thread tid:0x7fe7dbe15700 is not running
[2022-03-01 13:28:39] [INFO] the recovery thread was exited, reset tid
[2022-03-01 13:28:39] [NOTICE] Some nodes reconnect, all standby nodes are OK now
[2022-03-01 13:28:41] [NOTICE] new standby "node243" (ID: 1) has connected
[2022-03-01 13:31:31] [INFO] monitoring primary node "node248" (ID: 2) in normal state
7）查看备库数据库进程
8）原主库作为新备库rejoin到集群
[kingbase@node3 bin]$ ./repmgr cluster show
ID | Name | Role | Status | Upstream | Location | Priority | Timeline | Connection string
----+---------+---------+-----------+----------+----------+----------+----------+---------------------------
1 | node243 | standby | running | node248 | default | 100 | 5 | host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
2 | node248 | primary | * running | | default | 100 | 6 | host=192.168.7.248 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
9）主库查询流复制信息
test=# select * from sys_replication_slots;
slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn
---------------+--------+-----------+--------+----------+-----------+--------+------------+------+-------------
repmgr_slot_1 | | physical | | | f | t | 30928 | 1437 | | 0/120130A8 |
repmgr_slot_2 | | physical | | | f | f | | | | |
(2 rows)
test=# select * from sys_stat_replication;
pid | usesysid | usename | application_name | client_addr | client_hostname | client_port | backend_start | backend_xmin | state | sent_lsn | write_lsn | flush_lsn | replay_lsn | write_lag | flush_lag | replay_lag | sync_priority | sync_state | reply_time
-------+----------+---------+------------------+---------------+-----------------+-------------+-----------
30928 | 16384 | esrep | node243 | 192.168.7.243 | | 10817 | 2022-03-01 13:28:37.941077+08 | | streaming | 0/120130A8 | 0/120130A8 | 0/120130A8 | 0/120130A8 | | | | 1 | sync | 2022-03-01 13:32:08.445325+08
(1 row)
案例二：测试‘recovery = automatic’
1、查看集群节点状态信息：
[kingbase@node1 bin]$ ./repmgr cluster show
ID | Name | Role | Status | Upstream | Location | Priority | Timeline | Connection string
----+---------+---------+-----------+----------+----------+----------+----------+---------------------------
1 | node243 | primary | * running | | default | 100 | 3 | host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
2 | node248 | standby | running | node243 | default | 100 | 3 | host=192.168.7.248 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
2、配置recovery参数
[kingbase@node3 bin]$ cat ../etc/repmgr.conf |egrep -i 'recovery|failover'
failover='automatic'
recovery='automatic'
3、重启主库节点测试
[root@node3 ~]# reboot
4、查看备库hamgr日志
=如下所示，从日志中获知，主库节点宕机后，集群执行主备切换，并且在主库节点系统正常后，将原主库作为新备库自动加入到集群。=
[2022-03-01 14:38:09] [NOTICE] starting monitoring of node "node248" (ID: 2)
[2022-03-01 14:38:09] [INFO] "connection_check_type" set to "ping"
[2022-03-01 14:38:10] [INFO] monitoring connection to upstream node "node243" (ID: 1)
[2022-03-01 14:38:10] [NOTICE] try to change wal catched_up state to 1
[2022-03-01 14:38:10] [INFO] primary flush lsn is 0/17000578, local flush lsn is 0/170004C0
[2022-03-01 14:38:10] [NOTICE] try to change streaming_sync state to TRUE
[2022-03-01 14:43:11] [INFO] node "node248" (ID: 2) monitoring upstream node "node243" (ID: 1) in normal state
[2022-03-01 14:46:42] [WARNING] unable to ping "host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3"
[2022-03-01 14:46:42] [DETAIL] PQping() returned "PQPING_REJECT"
[2022-03-01 14:46:42] [WARNING] unable to connect to upstream node "node243" (ID: 1)
[2022-03-01 14:46:42] [INFO] sleeping 6 seconds until next reconnection attempt
[2022-03-01 14:46:48] [INFO] checking state of node 1, 1 of 10 attempts
[2022-03-01 14:46:58] [WARNING] unable to ping "user=esrep connect_timeout=10 dbname=esrep host=192.168.7.243 port=54321 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3 fallback_application_name=repmgr"
[2022-03-01 14:46:58] [DETAIL] PQping() returned "PQPING_NO_RESPONSE"
[2022-03-01 14:46:58] [INFO] sleeping 6 seconds until next reconnection attempt
......
[2022-03-01 14:48:59] [INFO] checking state of node 1, 10 of 10 attempts
[2022-03-01 14:48:59] [WARNING] unable to ping "user=esrep connect_timeout=10 dbname=esrep host=192.168.7.243 port=54321 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3 fallback_application_name=repmgr"
[2022-03-01 14:48:59] [DETAIL] PQping() returned "PQPING_NO_RESPONSE"
[2022-03-01 14:48:59] [WARNING] unable to reconnect to node 1 after 10 attempts
[2022-03-01 14:48:59] [NOTICE] setting "wal_retrieve_retry_interval" to 86405000 milliseconds
[2022-03-01 14:49:00] [WARNING] wal receiver not running
[2022-03-01 14:49:00] [NOTICE] WAL receiver disconnected on all sibling nodes
[2022-03-01 14:49:00] [INFO] WAL receiver disconnected on all 0 sibling nodes
[2022-03-01 14:49:00] [INFO] 0 active sibling nodes registered
[2022-03-01 14:49:00] [INFO] primary and this node have the same location ("default")
[2022-03-01 14:49:00] [INFO] no other sibling nodes - we win by default
[2022-03-01 14:49:00] [NOTICE] setting "wal_retrieve_retry_interval" to 5000 ms
[2022-03-01 14:49:00] [NOTICE] this node is the only available candidate and will now promote itself
[2022-03-01 14:49:00] [INFO] try to ping the trusted_servers "192.168.7.1" before execute promote_command
[2022-03-01 14:49:02] [NOTICE] PING 192.168.7.1 (192.168.7.1) 56(84) bytes of data.
--- 192.168.7.1 ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1002ms
rtt min/avg/max/mdev = 2.345/22.599/42.853/20.254 ms
[2022-03-01 14:49:02] [NOTICE] successfully ping one or more of the trusted_servers "192.168.7.1"
[2022-03-01 14:49:04] [NOTICE] PING 192.168.7.241 (192.168.7.241) 56(84) bytes of data.
--- 192.168.7.241 ping statistics ---
3 packets transmitted, 0 received, 100% packet loss, time 1999ms
[2022-03-01 14:49:04] [WARNING] ping host"192.168.7.241" failed
[2022-03-01 14:49:04] [DETAIL] average RTT value is not greater than zero
[2022-03-01 14:49:04] [INFO] loadvip result: 1, arping result: 1
[2022-03-01 14:49:04] [NOTICE] new primary node (ID: 2) acquire the virtual ip 192.168.7.241/24 success
[2022-03-01 14:49:04] [INFO] promote_command is:
"/home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/repmgr standby promote -f /home/kingbase/cluster/R6C5/R6C5R/kingbase/etc/repmgr.conf"
NOTICE: promoting standby to primary
DETAIL: promoting server "node248" (ID: 2) using sys_promote()
NOTICE: waiting up to 60 seconds (parameter "promote_check_timeout") for promotion to complete
INFO: SET synchronous TO "async" on primary host
[2022-03-01 14:49:07] [NOTICE] try to stop old primary db (host: "192.168.7.243")
NOTICE: STANDBY PROMOTE successful
DETAIL: server "node248" (ID: 2) was successfully promoted to primary
[2022-03-01 14:49:11] [INFO] switching to primary monitoring mode
[2022-03-01 14:49:11] [NOTICE] monitoring cluster primary "node248" (ID: 2)
[2022-03-01 14:49:11] [INFO] create a thread 0x7f1b4b125700 to check the cluster status
[2022-03-01 14:49:11] [INFO] child node: 1; attached: no
[2022-03-01 14:49:11] [INFO] check node status again, try 1 / 10 times
[2022-03-01 14:49:12] [INFO] node (ID: 1): no server running
.......
[2022-03-01 14:49:29] [INFO] check node status again, try 10 / 10 times
[2022-03-01 14:49:31] [INFO] child node: 1; attached: no
[2022-03-01 14:49:31] [INFO] found node down, recovery will be triggered after recovery delay time 20s
[2022-03-01 14:49:33] [INFO] child node: 1; attached: no
......
[2022-03-01 14:49:52] [INFO] child node: 1; attached: no
[2022-03-01 14:49:52] [INFO] recovery delay time reached. can do recovery now.
[2022-03-01 14:49:52] [INFO] [thread pid:11778] do_nodes_recovery thread begin. The pthread_t tid is 0x7f1b4b125700
[2022-03-01 14:49:52] [NOTICE] [thread pid:11778] node (ID: 1; host: "192.168.7.243") is not attached, ready to auto-recovery
[2022-03-01 14:49:52] [NOTICE] [thread pid:11778] Now, the primary host ip: 192.168.7.248
[2022-03-01 14:49:52] [INFO] [thread pid:11778] ES connection to host "192.168.7.243" succeeded, ready to do auto-recovery
[2022-03-01 14:49:53] [INFO] unlink file /tmp/.s.KINGBASE.54321.lock
[2022-03-01 14:49:53] [NOTICE] executing repmgr command "/home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/repmgr --dbname="host=192.168.7.248 dbname=esrep user=esrep port=54321" node rejoin --force-rewind"
NOTICE: sys_rewind execution required for this node to attach to rejoin target node 2
DETAIL: rejoin target server's timeline 8 forked off current database system timeline 7 before current recovery point 0/18000028
NOTICE: executing sys_rewind
DETAIL: sys_rewind command is "/home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/sys_rewind -D '/home/kingbase/cluster/R6C5/R6C5R/kingbase/data' --source-server='host=192.168.7.248 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3'"
sys_rewind: servers diverged at WAL location 0/17000680 on timeline 7
sys_rewind: rewinding from last common checkpoint at 0/160007C8 on timeline 7
sys_rewind: find last common checkpoint start time from 2022-03-01 14:49:53.170681 CST to 2022-03-01 14:49:53.296332 CST, in "0.125651" seconds.
sys_rewind: update the control file: minRecoveryPoint is '0/1700DE58', minRecoveryPointTLI is '8', and database state is 'in archive recovery'
sys_rewind: we will remove the dir '/home/kingbase/cluster/R6C5/R6C5R/kingbase/data/sys_replslot/repmgr_slot_2.rewind' and all the file/dir in it.
sys_rewind: rewind start wal location 0/16000798 (file 000000070000000000000016), end wal location 0/1700DE58 (file 000000080000000000000017). time from 2022-03-01 14:49:53.170681 CST to 2022-03-01 14:50:06.920859 CST, in "13.750178" seconds.
sys_rewind: Done!
NOTICE: 0 files copied to /home/kingbase/cluster/R6C5/R6C5R/kingbase/data
NOTICE: setting node 1's upstream to node 2
WARNING: unable to ping "host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3"
DETAIL: PQping() returned "PQPING_NO_RESPONSE"
NOTICE: begin to start server at 2022-03-01 14:50:07.530887
NOTICE: starting server using "/home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/sys_ctl -w -t 90 -D '/home/kingbase/cluster/R6C5/R6C5R/kingbase/data' -l /home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/logfile start"
NOTICE: start server finish at 2022-03-01 14:50:08.952996
NOTICE: NODE REJOIN successful
DETAIL: node 1 is now attached to node 2
[2022-03-01 14:50:09] [NOTICE] kbha: node (ID: 1) rejoin success.
[2022-03-01 14:50:10] [NOTICE] [thread pid:11778] node "node243" (ID: 1) auto-recovery success
[2022-03-01 14:50:10] [INFO] [thread pid:11778] do_nodes_recovery thread ends. The pthread_t tid is 0x7f1b4b125700
[2022-03-01 14:50:10] [INFO] SET synchronous TO "sync" on primary host
[2022-03-01 14:50:10] [INFO] thread tid:0x7f1b4b125700 is not running
[2022-03-01 14:50:10] [INFO] the recovery thread was exited, reset tid
[2022-03-01 14:50:10] [NOTICE] Some nodes reconnect, all standby nodes are OK now
[2022-03-01 14:50:12] [NOTICE] new standby "node243" (ID: 1) has connected
5、查看备库数据库进程和集群状态信息
[kingbase@node3 bin]$ ps -ef |grep kingbase
kingbase 2654 1 0 14:49 ? 00:00:00 /home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/kbha -A daemon -f /home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/../etc/repmgr.conf
kingbase 3462 1 0 14:50 ? 00:00:00 /home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/kingbase -D /home/kingbase/cluster/R6C5/R6C5R/kingbase/data
kingbase 3463 3462 0 14:50 ? 00:00:00 kingbase: logger
kingbase 3464 3462 0 14:50 ? 00:00:00 kingbase: startup recovering 000000080000000000000017
kingbase 3465 3462 0 14:50 ? 00:00:00 kingbase: checkpointer
kingbase 3466 3462 0 14:50 ? 00:00:00 kingbase: background writer
kingbase 3467 3462 0 14:50 ? 00:00:00 kingbase: stats collector
kingbase 3468 3462 0 14:50 ? 00:00:00 kingbase: walreceiver streaming 0/1700F160
kingbase 3471 3462 0 14:50 ? 00:00:00 kingbase: esrep esrep 192.168.7.243(57348) idle
kingbase 3522 1 0 14:50 ? 00:00:00 /home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/repmgrd -d -v -f /home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/../etc/repmgr.conf
kingbase 3523 3462 0 14:50 ? 00:00:00 kingbase: esrep esrep 192.168.7.243(57351) idle
[kingbase@node3 bin]$ ./repmgr cluster show
ID | Name | Role | Status | Upstream | Location | Priority | Timeline | Connection string
----+---------+---------+-----------+----------+----------+----------+----------+--------------------------
1 | node243 | standby | running | node248 | default | 100 | 7 | host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
2 | node248 | primary | * running | | default | 100 | 8 | host=192.168.7.248 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
案例三：测试‘recovery = manual’
1、查看集群节点状态信息：
[kingbase@node1 bin]$ ./repmgr cluster show
ID | Name | Role | Status | Upstream | Location | Priority | Timeline | Connection string
----+---------+---------+-----------+----------+----------+----------+----------+---------------------------
1 | node243 | primary | * running | | default | 100 | 3 | host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
2 | node248 | standby | running | node243 | default | 100 | 3 | host=192.168.7.248 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
2、查看recovery配置信息
3、重启主库主机系统
[root@node3 ~]# reboot
4、查看备库hamgr日志
=从以下日志信息获知，主库系统宕机后，集群执行主备切换，备库被提升为主库。==
[2022-03-02 10:32:38] [NOTICE] starting monitoring of node "node248" (ID: 2)
[2022-03-02 10:32:38] [INFO] "connection_check_type" set to "ping"
[2022-03-02 10:32:38] [INFO] monitoring connection to upstream node "node243" (ID: 1)
[2022-03-02 10:32:38] [NOTICE] try to change wal catched_up state to 1
[2022-03-02 10:32:38] [INFO] primary flush lsn is 0/1F000D40, local flush lsn is 0/1F000D40
[2022-03-02 10:32:38] [NOTICE] try to change streaming_sync state to TRUE
[2022-03-02 10:34:24] [WARNING] unable to ping "host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3"
[2022-03-02 10:34:24] [DETAIL] PQping() returned "PQPING_REJECT"
[2022-03-02 10:34:24] [WARNING] unable to connect to upstream node "node243" (ID: 1)
[2022-03-02 10:34:24] [INFO] sleeping 6 seconds until next reconnection attempt
[2022-03-02 10:34:30] [INFO] checking state of node 1, 1 of 10 attempts
[2022-03-02 10:34:40] [WARNING] unable to ping "user=esrep connect_timeout=10 dbname=esrep host=192.168.7.243 port=54321 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3 fallback_application_name=repmgr"
[2022-03-02 10:34:40] [DETAIL] PQping() returned "PQPING_NO_RESPONSE"
[2022-03-02 10:34:40] [INFO] sleeping 6 seconds until next reconnection attempt
......
[2022-03-02 10:35:47] [INFO] checking state of node 1, 10 of 10 attempts
[2022-03-02 10:35:47] [WARNING] unable to ping "user=esrep connect_timeout=10 dbname=esrep host=192.168.7.243 port=54321 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3 fallback_application_name=repmgr"
[2022-03-02 10:35:47] [DETAIL] PQping() returned "PQPING_NO_RESPONSE"
[2022-03-02 10:35:47] [WARNING] unable to reconnect to node 1 after 10 attempts
[2022-03-02 10:35:47] [NOTICE] setting "wal_retrieve_retry_interval" to 86405000 milliseconds
[2022-03-02 10:35:47] [WARNING] wal receiver not running
[2022-03-02 10:35:47] [NOTICE] WAL receiver disconnected on all sibling nodes
[2022-03-02 10:35:47] [INFO] WAL receiver disconnected on all 0 sibling nodes
[2022-03-02 10:35:47] [INFO] 0 active sibling nodes registered
[2022-03-02 10:35:47] [INFO] primary and this node have the same location ("default")
[2022-03-02 10:35:47] [INFO] no other sibling nodes - we win by default
[2022-03-02 10:35:47] [NOTICE] setting "wal_retrieve_retry_interval" to 5000 ms
[2022-03-02 10:35:48] [NOTICE] this node is the only available candidate and will now promote itself
[2022-03-02 10:35:48] [INFO] try to ping the trusted_servers "192.168.7.1" before execute promote_command
[2022-03-02 10:35:50] [NOTICE] PING 192.168.7.1 (192.168.7.1) 56(84) bytes of data.
--- 192.168.7.1 ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1008ms
rtt min/avg/max/mdev = 2.473/2.535/2.598/0.080 ms
[2022-03-02 10:35:50] [NOTICE] successfully ping one or more of the trusted_servers "192.168.7.1"
[2022-03-02 10:35:51] [NOTICE] PING 192.168.7.241 (192.168.7.241) 56(84) bytes of data.
--- 192.168.7.241 ping statistics ---
2 packets transmitted, 0 received, +1 errors, 100% packet loss, time 1000ms
[2022-03-02 10:35:51] [WARNING] ping host"192.168.7.241" failed
[2022-03-02 10:35:51] [DETAIL] average RTT value is not greater than zero
[2022-03-02 10:35:51] [INFO] loadvip result: 1, arping result: 1
[2022-03-02 10:35:51] [NOTICE] new primary node (ID: 2) acquire the virtual ip 192.168.7.241/24 success
[2022-03-02 10:35:51] [INFO] promote_command is:
"/home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/repmgr standby promote -f /home/kingbase/cluster/R6C5/R6C5R/kingbase/etc/repmgr.conf"
NOTICE: promoting standby to primary
DETAIL: promoting server "node248" (ID: 2) using sys_promote()
NOTICE: waiting up to 60 seconds (parameter "promote_check_timeout") for promotion to complete
[2022-03-02 10:35:51] [NOTICE] try to stop old primary db (host: "192.168.7.243")
INFO: SET synchronous TO "async" on primary host
NOTICE: STANDBY PROMOTE successful
DETAIL: server "node248" (ID: 2) was successfully promoted to primary
[2022-03-02 10:35:56] [INFO] 0 followers to notify
[2022-03-02 10:35:56] [INFO] switching to primary monitoring mode
[2022-03-02 10:35:56] [NOTICE] monitoring cluster primary "node248" (ID: 2)
[2022-03-02 10:35:56] [INFO] create a thread 0x7fdeaa4b9700 to check the cluster status
[2022-03-02 10:35:57] [INFO] node (ID: 1): no server running
[2022-03-02 10:35:57] [INFO] [thread 0x7fdeaa4b9700] the cluster has no other running primary node, exit
5、原主库系统正常启动
1）从新主库查看集群状态 信息
=从以下信息可以获知，集群现在处于‘双主’状态，只是原主库是‘failed’，无法连接。=
[kingbase@node1 bin]$ ./repmgr cluster show
ID | Name | Role | Status | Upstream | Location | Priority | Timeline | Connection string
----+---------+---------+-----------+----------+----------+----------+----------+----------------
1 | node243 | primary | - failed | | default | 100 | ? | host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
2 | node248 | primary | * running | | default | 100 | 10 | host=192.168.7.248 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
WARNING: following issues were detected
- unable to connect to node "node243" (ID: 1)
You have new mail in /var/spool/mail/kingbase
2）在新主库（原备库）创建复制槽
# 创建replication slots
test=# select sys_create_physical_replication_slot('repmgr_slot_1');
sys_create_physical_replication_slot
--------------------------------------
(repmgr_slot_1,)
(1 row)
test=# select sys_create_physical_replication_slot('repmgr_slot_2');
sys_create_physical_replication_slot
--------------------------------------
(repmgr_slot_2,)
(1 row)
test=# select * from sys_replication_slots;
slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn
---------------+--------+-----------+--------+----------+-----------+--------+------------+-----
repmgr_slot_1 | | physical | | | f | f | | | | |
repmgr_slot_2 | | physical | | | f | f | | | | |
(2 rows)
3）在原主库（新主库）执行以下恢复操作
# 备份data目录
[kingbase@node3 kingbase]$ cp data data.bk -r
# 生成备库标识文件
[kingbase@node3 kingbase]$ cd data
[kingbase@node3 data]$ touch standby.signal
4）在原主库执行repmgr node rejoin重新加入到集群
[kingbase@node3 bin]$ ./repmgr node rejoin -h 192.168.7.248 -U esrep -d esrep --force-rewind
NOTICE: sys_rewind execution required for this node to attach to rejoin target node 2
DETAIL: rejoin target server's timeline 10 forked off current database system timeline 9 before current recovery point 0/200000A0
NOTICE: executing sys_rewind
DETAIL: sys_rewind command is "/home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/sys_rewind -D '/home/kingbase/cluster/R6C5/R6C5R/kingbase/data' --source-server='host=192.168.7.248 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3'"
sys_rewind: servers diverged at WAL location 0/1F000D70 on timeline 9
sys_rewind: rewinding from last common checkpoint at 0/1E000A70 on timeline 9
sys_rewind: find last common checkpoint start time from 2022-03-02 10:52:34.133058 CST to 2022-03-02 10:52:34.358066 CST, in "0.225008" seconds.
sys_rewind: update the control file: minRecoveryPoint is '0/1F011AD0', minRecoveryPointTLI is '10', and database state is 'in archive recovery'
sys_rewind: rewind start wal location 0/1E000A40 (file 00000009000000000000001E), end wal location 0/1F011AD0 (file 0000000A000000000000001F). time from 2022-03-02 10:52:34.133058 CST to 2022-03-02 10:53:06.442270 CST, in "32.309212" seconds.
sys_rewind: Done!
NOTICE: 0 files copied to /home/kingbase/cluster/R6C5/R6C5R/kingbase/data
NOTICE: setting node 1's upstream to node 2
WARNING: unable to ping "host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3"
DETAIL: PQping() returned "PQPING_NO_RESPONSE"
NOTICE: begin to start server at 2022-03-02 10:53:06.588331
NOTICE: starting server using "/home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/sys_ctl -w -t 90 -D '/home/kingbase/cluster/R6C5/R6C5R/kingbase/data' -l /home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/logfile start"
NOTICE: start server finish at 2022-03-02 10:53:07.313294
NOTICE: NODE REJOIN successful
DETAIL: node 1 is now attached to node 2
5）启动新备库数据库服务
[kingbase@node3 bin]$ ps -ef |grep kingbase
kingbase 3218 1 0 10:36 ? 00:00:00 /home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/kbha -A daemon -f /home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/../etc/repmgr.conf
kingbase 5817 1 0 10:49 ? 00:00:01 /home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/repmgrd -d -v -f /home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/../etc/repmgr.conf
kingbase 6730 1 0 10:53 ? 00:00:00 /home/kingbase/cluster/R6C5/R6C5R/kingbase/bin/kingbase -D /home/kingbase/cluster/R6C5/R6C5R/kingbase/data
kingbase 6731 6730 0 10:53 ? 00:00:00 kingbase: logger
kingbase 6732 6730 0 10:53 ? 00:00:00 kingbase: startup recovering 0000000A000000000000001F
kingbase 6736 6730 0 10:53 ? 00:00:00 kingbase: checkpointer
kingbase 6737 6730 0 10:53 ? 00:00:00 kingbase: background writer
kingbase 6738 6730 0 10:53 ? 00:00:00 kingbase: stats collector
kingbase 6739 6730 0 10:53 ? 00:00:00 kingbase: walreceiver streaming 0/1F012A78
kingbase 6743 6730 0 10:53 ? 00:00:00 kingbase: esrep esrep 192.168.7.243(55941) idle
kingbase 6750 6730 0 10:53 ? 00:00:00 kingbase: esrep esrep 192.168.7.243(55947) idle
6）查看集群节点状态
[kingbase@node3 bin]$ ./repmgr cluster show
ID | Name | Role | Status | Upstream | Location | Priority | Timeline | Connection string
----+---------+---------+-----------+----------+----------+----------+----------+----------------
1 | node243 | standby | running | node248 | default | 100 | 9 | host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
2 | node248 | primary | * running | | default | 100 | 10 | host=192.168.7.248 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
7）重启集群测试（可选）
[kingbase@node3 bin]$ ./sys_monitor.sh restart
2022-03-02 10:55:26 Ready to stop all DB ...
....
server started
2022-03-02 10:55:52 execute to start DB on "[192.168.7.248]" success, connect to check it.
2022-03-02 10:55:53 DB on "[192.168.7.248]" start success.
ID | Name | Role | Status | Upstream | Location | Priority | Timeline | Connection string
----+---------+---------+-----------+-----------+----------+----------+----------+---------------
1 | node243 | standby | running | ! node248 | default | 100 | 10 | host=192.168.7.243 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
2 | node248 | primary | * running | | default | 100 | 10 | host=192.168.7.248 user=esrep dbname=esrep port=54321 connect_timeout=10 keepalives=1 keepalives_idle=10 keepalives_interval=1 keepalives_count=3
WARNING: following issues were detected
- node "node243" (ID: 1) is not attached to its upstream node "node248" (ID: 2)
2022-03-02 10:55:53 The primary DB is started.
......
2022-03-02 10:56:15 repmgrd on "[192.168.7.248]" start success.
ID | Name | Role | Status | Upstream | repmgrd | PID | Paused? | Upstream last seen
----+---------+---------+-----------+----------+---------+-------+---------+--------------------
1 | node243 | standby | running | node248 | running | 9500 | no | 1 second(s) ago
2 | node248 | primary | * running | | running | 27881 | no | n/a
[2022-03-02 10:56:18] [NOTICE] redirecting logging output to "/home/kingbase/cluster/R6C5/R6C5R/kingbase/log/kbha.log"
[2022-03-02 10:56:20] [NOTICE] redirecting logging output to "/home/kingbase/cluster/R6C5/R6C5R/kingbase/log/kbha.log"
2022-03-02 10:56:22 Done.
=从以上信息获知，通过手工执行repmgr node rejoin，原主库作为新备库重新加入到集群中。=
总结：
1、对于recovery=standby，主库节点系统宕机后，集群执行主库切换，原主库需要人工配置为备库模式，并启动数据库服务，然后集群可自动将其加入到集群。
2、对于recovery=automatic，主库节点系统宕机后，集群执行主库切换，不需要人工参与，原主库将作为新的备库自动加入到集群。
3、对于recovery=manual，主库节点系统宕机后，集群执行主库切换，需要人工参与，在原主库执行‘repmgr node rejoin’操作，将原主库将作为新的备库自动加入到集群。
4、对于无DBA日常监控管理的生产环境，可以考虑将recovery配置为automatic，提升集群架构的可靠性。
KINGBASE研究院
