MySQL--MHA与GTID

##==========================================##

MySQL 5.6版本引入GTID来解决主从切换时BINLOG位置点难定位的问题，MHA从0.56版本开始支持基于GTID的复制，在发生故障切换时判断群集是否能采用基于GTID的方式进行切换

##==========================================##
基于GTID进行故障切换的条件：
1、所有节点开启GTID模式，设置gtid_mode=1
2、所有节点上Executed_Gtid_Set不为空
3、至少一个节点使用Auto_Position=1

##==========================================##
基于GTID进行故障切换：
1、如果候选Master节点不拥有最新的Relay log，那么将候选Master连接到拥有最新Relay log的Salve上进行日志补偿
2、如果群集中使用Binlog Server，则尝试从Binlog Server上拉取缺失的Binlog并应用到候选Master上
3、候选Matser拥有最新数据，将其升级为新Master，将其他slave连接到新Master上进行数据同步，可以给masterha_master_switch传入–wait_until_gtid_in_sync=1参数使其不等其它Slave完成数据同步，以加快切换速度。

##==========================================##
基于GTID模式进行故障切换时，无论原Master节点OS是否正常，都不会尝试从原Master节点读取BINLOG进行日志补偿。
基于GTID模式的MHA支持在复制拓扑中使用BINLOG Server来进行日志补偿，而非GTID模式的MHA会忽略BINLOG Server。
建议在基于GTID模式的群集中，不使用MHA进行"手动主从切换"，该操作可能会导致原主库上部分BINLOG丢失。

##==========================================##
在非GTID模式下，会先进行Phase 3.1阶段，从拥有最新BINLOG的从库上获取差异日志，再进行Phase 3.2阶段，尝试从原Master服务器上获取最新BINLOG。
 

使用非GTID模式切换的日志

##==========================================##
在基于GTID模式下，不会进行Phase 3.2阶段，即尝试从原Master服务器中获取最新BINLOG。

使用GTID模式切换的日志：

Sun Jul   ::  - [info] MHA::MasterMonitor version 0.56.
Sun Jul   ::  - [info] GTID failover mode =
Sun Jul   ::  - [info] Dead Servers:
Sun Jul   ::  - [info] Alive Servers:
Sun Jul   ::  - [info]   10.0.203.104(10.0.203.104:)
Sun Jul   ::  - [info]   10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info]   10.0.203.117(10.0.203.117:)
Sun Jul   ::  - [info] Alive Slaves:
Sun Jul   ::  - [info]   10.0.203.104(10.0.203.104:)  Version=5.7.-log (oldest major version between slaves) log-bin:enabled
Sun Jul   ::  - [info]     GTID ON
Sun Jul   ::  - [info]     Replicating from 10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info]     Primary candidate for the new Master (candidate_master is set)
Sun Jul   ::  - [info]   10.0.203.117(10.0.203.117:)  Version=5.7.-log (oldest major version between slaves) log-bin:enabled
Sun Jul   ::  - [info]     GTID ON
Sun Jul   ::  - [info]     Replicating from 10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info] Current Alive Master: 10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info] Checking slave configurations..
Sun Jul   ::  - [info]  read_only= is not set on slave 10.0.203.104(10.0.203.104:).
Sun Jul   ::  - [info]  read_only= is not set on slave 10.0.203.117(10.0.203.117:).
Sun Jul   ::  - [info] Checking replication filtering settings..
Sun Jul   ::  - [info]  binlog_do_db= , binlog_ignore_db=
Sun Jul   ::  - [info]  Replication filtering check ok.
Sun Jul   ::  - [info] GTID (with auto-pos) is supported. Skipping all SSH and Node package checking.
Sun Jul   ::  - [info] Checking SSH publickey authentication settings on the current master..
Sun Jul   ::  - [info] HealthCheck: SSH to 10.0.203.109 is reachable.
Sun Jul   ::  - [info]
10.0.203.109(10.0.203.109:) (current master)
 +--10.0.203.104(10.0.203.104:)
 +--10.0.203.117(10.0.203.117:)

Sun Jul   ::  - [warning] master_ip_failover_script is not defined.
Sun Jul   ::  - [warning] shutdown_script is not defined.
Sun Jul   ::  - [info] Set master ping interval  seconds.
Sun Jul   ::  - [warning] secondary_check_script is not defined. It is highly recommended setting it to check master reachability from two or more routes.
Sun Jul   ::  - [info] Starting ping health check on 10.0.203.109(10.0.203.109:)..
Sun Jul   ::  - [info] Ping(SELECT) succeeded, waiting until MySQL doesn't respond..
Sun Jul   ::  - [warning] Got error on MySQL select ping:  (MySQL server has gone away)
Sun Jul   ::  - [info] Executing SSH check script: exit
Sun Jul   ::  - [info] HealthCheck: SSH to 10.0.203.109 is reachable.
Sun Jul   ::  - [warning] Got error on MySQL connect:  (Lost connection to MySQL server at 'reading initial communication packet', system error: )
Sun Jul   ::  - [warning] Connection failed  time(s)..
Sun Jul   ::  - [warning] Got error on MySQL connect:  (Lost connection to MySQL server at 'reading initial communication packet', system error: )
Sun Jul   ::  - [warning] Connection failed  time(s)..
Sun Jul   ::  - [warning] Got error on MySQL connect:  (Lost connection to MySQL server at 'reading initial communication packet', system error: )
Sun Jul   ::  - [warning] Connection failed  time(s)..
Sun Jul   ::  - [warning] Master is not reachable from health checker!
Sun Jul   ::  - [warning] Master 10.0.203.109(10.0.203.109:) is not reachable!
Sun Jul   ::  - [warning] SSH is reachable.
Sun Jul   ::  - [info] Connecting to a master server failed. Reading configuration file /etc/masterha_default.cnf and /etc/masterha/app1.cnf again, and trying to connect to all servers to check server status..
Sun Jul   ::  - [warning] Global configuration file /etc/masterha_default.cnf not found. Skipping.
Sun Jul   ::  - [info] Reading application default configuration from /etc/masterha/app1.cnf..
Sun Jul   ::  - [info] Reading server configuration from /etc/masterha/app1.cnf..
Sun Jul   ::  - [info] GTID failover mode =
Sun Jul   ::  - [info] Dead Servers:
Sun Jul   ::  - [info]   10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info] Alive Servers:
Sun Jul   ::  - [info]   10.0.203.104(10.0.203.104:)
Sun Jul   ::  - [info]   10.0.203.117(10.0.203.117:)
Sun Jul   ::  - [info] Alive Slaves:
Sun Jul   ::  - [info]   10.0.203.104(10.0.203.104:)  Version=5.7.-log (oldest major version between slaves) log-bin:enabled
Sun Jul   ::  - [info]     GTID ON
Sun Jul   ::  - [info]     Replicating from 10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info]     Primary candidate for the new Master (candidate_master is set)
Sun Jul   ::  - [info]   10.0.203.117(10.0.203.117:)  Version=5.7.-log (oldest major version between slaves) log-bin:enabled
Sun Jul   ::  - [info]     GTID ON
Sun Jul   ::  - [info]     Replicating from 10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info] Checking slave configurations..
Sun Jul   ::  - [info]  read_only= is not set on slave 10.0.203.104(10.0.203.104:).
Sun Jul   ::  - [info]  read_only= is not set on slave 10.0.203.117(10.0.203.117:).
Sun Jul   ::  - [info] Checking replication filtering settings..
Sun Jul   ::  - [info]  Replication filtering check ok.
Sun Jul   ::  - [info] Master is down!
Sun Jul   ::  - [info] Terminating monitoring script.
Sun Jul   ::  - [info] Got exit code  (Master dead).
Sun Jul   ::  - [info] MHA::MasterFailover version 0.56.
Sun Jul   ::  - [info] Starting master failover.
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] * Phase : Configuration Check Phase..
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] GTID failover mode =
Sun Jul   ::  - [info] Dead Servers:
Sun Jul   ::  - [info]   10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info] Checking master reachability via MySQL(double check)...
Sun Jul   ::  - [info]  ok.
Sun Jul   ::  - [info] Alive Servers:
Sun Jul   ::  - [info]   10.0.203.104(10.0.203.104:)
Sun Jul   ::  - [info]   10.0.203.117(10.0.203.117:)
Sun Jul   ::  - [info] Alive Slaves:
Sun Jul   ::  - [info]   10.0.203.104(10.0.203.104:)  Version=5.7.-log (oldest major version between slaves) log-bin:enabled
Sun Jul   ::  - [info]     GTID ON
Sun Jul   ::  - [info]     Replicating from 10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info]     Primary candidate for the new Master (candidate_master is set)
Sun Jul   ::  - [info]   10.0.203.117(10.0.203.117:)  Version=5.7.-log (oldest major version between slaves) log-bin:enabled
Sun Jul   ::  - [info]     GTID ON
Sun Jul   ::  - [info]     Replicating from 10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info] Starting GTID based failover.
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] ** Phase : Configuration Check Phase completed.
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] * Phase : Dead Master Shutdown Phase..
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] Forcing shutdown so that applications never connect to the current master..
Sun Jul   ::  - [warning] master_ip_failover_script is not set. Skipping invalidating dead master IP address.
Sun Jul   ::  - [warning] shutdown_script is not set. Skipping explicit shutting down of the dead master.
Sun Jul   ::  - [info] * Phase : Dead Master Shutdown Phase completed.
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] * Phase : Master Recovery Phase..
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] * Phase 3.1: Getting Latest Slaves Phase..
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] The latest binary log file/position on all slaves is mysql-bin.:
Sun Jul   ::  - [info] Retrieved Gtid Set: 541e0f07--11e8--0800270b00d2:-
Sun Jul   ::  - [info] Latest slaves (Slaves that received relay log files to the latest):
Sun Jul   ::  - [info]   10.0.203.104(10.0.203.104:)  Version=5.7.-log (oldest major version between slaves) log-bin:enabled
Sun Jul   ::  - [info]     GTID ON
Sun Jul   ::  - [info]     Replicating from 10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info]     Primary candidate for the new Master (candidate_master is set)
Sun Jul   ::  - [info]   10.0.203.117(10.0.203.117:)  Version=5.7.-log (oldest major version between slaves) log-bin:enabled
Sun Jul   ::  - [info]     GTID ON
Sun Jul   ::  - [info]     Replicating from 10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info] The oldest binary log file/position on all slaves is mysql-bin.:
Sun Jul   ::  - [info] Retrieved Gtid Set: 541e0f07--11e8--0800270b00d2:-
Sun Jul   ::  - [info] Oldest slaves:
Sun Jul   ::  - [info]   10.0.203.104(10.0.203.104:)  Version=5.7.-log (oldest major version between slaves) log-bin:enabled
Sun Jul   ::  - [info]     GTID ON
Sun Jul   ::  - [info]     Replicating from 10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info]     Primary candidate for the new Master (candidate_master is set)
Sun Jul   ::  - [info]   10.0.203.117(10.0.203.117:)  Version=5.7.-log (oldest major version between slaves) log-bin:enabled
Sun Jul   ::  - [info]     GTID ON
Sun Jul   ::  - [info]     Replicating from 10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] * Phase 3.3: Determining New Master Phase..
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] Searching new master from slaves..
Sun Jul   ::  - [info]  Candidate masters from the configuration file:
Sun Jul   ::  - [info]   10.0.203.104(10.0.203.104:)  Version=5.7.-log (oldest major version between slaves) log-bin:enabled
Sun Jul   ::  - [info]     GTID ON
Sun Jul   ::  - [info]     Replicating from 10.0.203.109(10.0.203.109:)
Sun Jul   ::  - [info]     Primary candidate for the new Master (candidate_master is set)
Sun Jul   ::  - [info]  Non-candidate masters:
Sun Jul   ::  - [info]  Searching from candidate_master slaves which have received the latest relay log events..
Sun Jul   ::  - [info] New master is 10.0.203.104(10.0.203.104:)
Sun Jul   ::  - [info] Starting master failover..
Sun Jul   ::  - [info]
From:
10.0.203.109(10.0.203.109:) (current master)
 +--10.0.203.104(10.0.203.104:)
 +--10.0.203.117(10.0.203.117:)

To:
10.0.203.104(10.0.203.104:) (new master)
 +--10.0.203.117(10.0.203.117:)
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] * Phase 3.3: New Master Recovery Phase..
Sun Jul   ::  - [info]
Sun Jul   ::  - [info]  Waiting all logs to be applied..
Sun Jul   ::  - [info]   done.
Sun Jul   ::  - [info] Getting new master's binlog name and position..
Sun Jul   ::  - [info]  mysql-bin.:
Sun Jul   ::  - [info]  All other slaves should start replication from here. Statement should be: CHANGE MASTER TO MASTER_HOST='10.0.203.104', MASTER_PORT=, MASTER_AUTO_POSITION=, MASTER_USER='replicater', MASTER_PASSWORD='xxx';
Sun Jul   ::  - [info] Master Recovery succeeded. File:Pos:Exec_Gtid_Set: mysql-bin., , 41d8a420--11e8--080027e837eb:-,
541e0f07--11e8--0800270b00d2:-
Sun Jul   ::  - [warning] master_ip_failover_script is not set. Skipping taking over new master IP address.
Sun Jul   ::  - [info] ** Finished master recovery successfully.
Sun Jul   ::  - [info] * Phase : Master Recovery Phase completed.
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] * Phase : Slaves Recovery Phase..
Sun Jul   ::  - [info]
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] * Phase 4.1: Starting Slaves in parallel..
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] -- Slave recovery on host 10.0.203.117(10.0.203.117:) started, pid: . Check tmp log /var/log/masterha/app1/10.0..117_3358_20180708233601.log if it takes time..
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] Log messages from 10.0.203.117 ...
Sun Jul   ::  - [info]
Sun Jul   ::  - [info]  Resetting slave 10.0.203.117(10.0.203.117:) and starting replication from the new master 10.0.203.104(10.0.203.104:)..
Sun Jul   ::  - [info]  Executed CHANGE MASTER.
Sun Jul   ::  - [info]  Slave started.
Sun Jul   ::  - [info]  gtid_wait(41d8a420--11e8--080027e837eb:-,
541e0f07--11e8--0800270b00d2:-) completed on 10.0.203.117(10.0.203.117:). Executed  events.
Sun Jul   ::  - [info] End of log messages from 10.0.203.117.
Sun Jul   ::  - [info] -- Slave on host 10.0.203.117(10.0.203.117:) started.
Sun Jul   ::  - [info] All new slave servers recovered successfully.
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] * Phase : New master cleanup phase..
Sun Jul   ::  - [info]
Sun Jul   ::  - [info] Resetting slave info on the new master..
Sun Jul   ::  - [info]  10.0.203.104: Resetting slave info succeeded.
Sun Jul   ::  - [info] Master failover to 10.0.203.104(10.0.203.104:) completed successfully.
Sun Jul   ::  - [info] 

----- Failover Report -----

app1: MySQL Master failover 10.0.203.109(10.0.203.109:) to 10.0.203.104(10.0.203.104:) succeeded

Master 10.0.203.109(10.0.203.109:) is down!

Check MHA Manager logs at localhost.localdomain:/var/log/masterha/app1/manager.log for details.

Started automated(non-interactive) failover.
Selected 10.0.203.104(10.0.203.104:) as a new master.
10.0.203.104(10.0.203.104:): OK: Applying all logs succeeded.
10.0.203.117(10.0.203.117:): OK: Slave started, replicating from 10.0.203.104(10.0.203.104:)
10.0.203.104(10.0.203.104:): Resetting slave info succeeded.
Master failover to 10.0.203.104(10.0.203.104:) completed successfully.

##==========================================##
MHA在检查时，使用SHOW SLAVE STATUS获取结构中Auto_Position的值来判断是否使用master_auto_position参数来搭建主从复制。

MHA在切换时，如果使用非GTID模式切换，则在CHANGE MASTER中不会带上参数master_auto_position=0，而如果该从库之前配置为master_auto_position=1，那么CHANGE MASTER会报错，无法正常进行切换。

因此不能简单修改Server.PM或DBHelper.pm文件来将基于GTID模式切换的群集修改为使用非GTID模式进行切换。

##==========================================##

如果群集因为某种原因导致主从节点上的Executed_Gtid_Set不同，如：

1、对从库进行直接授权，导致从库比主库拥有更多BINLOG，但该Binlog因各种原因被Purged掉

2、群集做过版本升级，从未使用GTID的版本升级到GTID版本，从库上曾一段时间内作为主库提供服务，但该时间段日志被Purged掉

有上诉类似问题时，将从库提升为主库并使用master_auto_position=1来配置复制，复制会因为新主库无法提供足够BINLOG事件而失败。

处理办法：

1、通过RESET MASTER和SET GLOBAL gtid_purged=''使得所有节点拥有相同的GTID 集合

2、将所有复制修改为基于POS点搭建的复制。

##==========================================##