从错误日志里的提示：《target master‘s advisory lock is already held by someone. Please check whether you monitor the same master。。。》

mysql主节点是否存在。

mha代码里内容：也是验证主库状态

    ( $ret, $ssh_reachable ) = $master_ping->wait_until_unreachable();
    ( $ret, $ssh_reachable ) = $_master_ping->wait_until_unreachable();
    if ( $ret eq '2' ) {
      $log->error(
"Target master's advisory lock is already held by someone. Please check whether you monitor the same master from multiple monitoring processes."

杀死对应监控集群主库MHA连接的会话( kill xxx; )或者重启主库

• 杀死对应监控集群主库MHA连接的会话即可，原理
  
  • MHA在监控对应实例之前，会发起一个锁请求SELECT GET_LOCK('MHA_Master_High_Availability_Monitor', ?) AS Value，如果加锁成功则返回成功，如果加锁不成功则会提示有另外的进程在监控该实例
  • MHA在结束监控之后，会释放锁SELECT RELEASE_LOCK('MHA_Master_High_Availability_Monitor') As Value，或者直接关闭连接（也可以释放锁）
  • 如果MHA在异常退出之后没有关闭MySQL的会话，那么该锁会一直存在，且无法再次启动MHA进程
• 其他过程加锁

use constant Get_Failover_Lock_SQL =>

  "SELECT GET_LOCK('MHA_Master_High_Availability_Failover', ?) AS Value";

use constant Release_Failover_Lock_SQL =>

  "SELECT RELEASE_LOCK('MHA_Master_High_Availability_Failover') As Value";

use constant Get_Monitor_Lock_SQL =>

  "SELECT GET_LOCK('MHA_Master_High_Availability_Monitor', ?) AS Value";

use constant Release_Monitor_Lock_SQL =>

  "SELECT RELEASE_LOCK('MHA_Master_High_Availability_Monitor') As Value";