因CM不满足多数派原则导致集群无法使用的问题
一、问题现象
对于节点个数大于或等于3的集群来说。如果利用cm_ctl stop -n 2...N命令停止多节点，只留有单节点存活那么集群将无法正常使用，且会在如下两种状态反复切换：
状态一：
[ CMServer State ]
node node_ip instance state
------------------------------------------------------------------------------------------------
1 openGauss151 NodeIp_1 1 /data/ss_development_install_1/install/cm/cm_server Primary
2 openGauss153 NodeIp_2 2 /data/ss_development_install_1/install/cm/cm_server Down
3 openGauss155 NodeIp_3 3 /data/ss_development_install_1/install/cm/cm_server Down
[ Defined Resource State ]
node node_ip res_name instance state
----------------------------------------------------------
1 openGauss151 NodeIp_1 dms_res 6001 OnLine
2 openGauss153 NodeIp_2 dms_res 6002 Deleted
3 openGauss155 NodeIp_3 dms_res 6003 Deleted
1 openGauss151 NodeIp_1 dss 20001 OnLine
2 openGauss153 NodeIp_2 dss 20002 Deleted
3 openGauss155 NodeIp_3 dss 20003 Deleted
[ Cluster State ]
cluster_state : Degraded
redistributing : No
balanced : Yes
current_az : AZ_ALL
[ Datanode State ]
node node_ip instance state
--------------------------------------------------------------------------------------------------------------
1 openGauss151 NodeIp_1 6001 22290 /data/ss_development_install_1/install/data/dn P Primary Normal
2 openGauss153 NodeIp_2 6002 22290 /data/ss_development_install_1/install/data/dn S Down Unknown
3 openGauss155 NodeIp_3 6003 22290 /data/ss_development_install_1/install/data/dn S Down Unknown
状态二：
[ CMServer State ]
node node_ip instance state
------------------------------------------------------------------------------------------------
1 openGauss151 NodeIp_1 1 /data/ss_development_install_1/install/cm/cm_server Standby
2 openGauss153 NodeIp_2 2 /data/ss_development_install_1/install/cm/cm_server Down
3 openGauss155 NodeIp_3 3 /data/ss_development_install_1/install/cm/cm_server Down
cm_ctl: can't connect to cm_server.
Maybe cm_server is not running, or timeout expired. Please try again.
二、根因分析
出现上述状态是由于CMS主节点是基于抢锁机制来选主的，与此同时CMS还有另外一套检测机制，如果是大于等于3节点的集群环境，若发现连接CMS主节点的cm_agent进程小于等于1那么会强制将当前的CMS主降级为备（这种设计本质上是防止脑裂场景），但是抢锁线程发现当前集群不存在CMS主节点，又会主动去抢锁，以至于集群会在两种状态切换。总体来说，当前集群不可用。
三、 解决方案
出现该情况，并没有直接的解决方案，若想集群恢复，只能利用cm_ctl start -n 1...N命令启动其他节点，以保证集群不再是单节点运行。
意见反馈