系统变量group_replication_group_seeds为空导致MySQL节点无法启动组复制

MySQL InnoDB Cluster集群中一个节点,在服务器重启过后,启动MySQL实例后,发现status为MISSING,另外memberState为OFFLINE状态。如下所示:

 MySQL  mysqldbu02:7306 ssl  JS > cluster.status()
{
    "clusterName""yssps"
    "defaultReplicaSet": {
        "name""default"
        "primary""mysqldbu02:7306"
        "ssl""REQUIRED"
        "status""OK_NO_TOLERANCE_PARTIAL"
        "statusText""Cluster is NOT tolerant to any failures. 1 member is not active."
        "topology": {
            "mysqldbu01:7306": {
                "address""mysqldbu01:7306"
                "instanceErrors": [
                    "NOTE: group_replication is stopped."
                ], 
                "memberRole""SECONDARY"
                "memberState""OFFLINE"
                "mode""n/a"
                "readReplicas": {}, 
                "role""HA"
                "status""(MISSING)"
                "version""8.0.35"
            }, 
            "mysqldbu02:7306": {
                "address""mysqldbu02:7306"
                "memberRole""PRIMARY"
                "mode""R/W"
                "readReplicas": {}, 
                "replicationLag""applier_queue_applied"
                "role""HA"
                "status""ONLINE"
                "version""8.0.35"
            }, 
            "mysqldbu03:7306": {
                "address""mysqldbu03:7306"
                "memberRole""SECONDARY"
                "mode""R/O"
                "readReplicas": {}, 
                "replicationLag""applier_queue_applied"
                "role""HA"
                "status""ONLINE"
                "version""8.0.35"
            }
        }, 
        "topologyMode""Single-Primary"
    }, 
    "groupInformationSourceMember""mysqldbu02:7306"
}
 MySQL  mysqldbu02:7306 ssl  JS > 

从错误提示信息来看是组复制没有启动(group_replication is stopped)。其实MySQL InnoDB Cluster的节点重启操作在生产环境和测试环境都做过多次,还是第一次遇到这种情况。

另外,在这个节点上查看节点的状态如下:

mysql> select * from performance_schema.replication_group_members order by member_host\G
*************************** 1. row ***************************
              CHANNEL_NAME: group_replication_applier
                 MEMBER_ID: 5ad73846-3fce-11ee-a86c-5254003c1c6e
               MEMBER_HOST: mysqldbu01
               MEMBER_PORT: 7306
              MEMBER_STATE: OFFLINE
               MEMBER_ROLE: 
            MEMBER_VERSION: 
MEMBER_COMMUNICATION_STACK: MySQL
1 row in set (0.00 sec)

mysql>

具体的错误日志如下所示

2024-11-06T10:03:19.369090+08:00 29 [Warning] [MY-011735] [Repl] Plugin group_replication reported: '[GCS] Automatically adding IPv4
 localhost address to the allowlist. It is mandatory that it is added.'

2024-11-06T10:03:19.369171+08:00 29 [Warning] [MY-011735] [Repl] Plugin group_replication reported: '[GCS] Automatically adding IPv6
 localhost address to the allowlist. It is mandatory that it is added.'

2024-11-06T10:03:19.371586+08:00 40 [System] [MY-010597] [Repl] 'CHANGE REPLICATION SOURCE TO FOR CHANNEL 'group_replication_applier
' executed'. Previous state source_host='<NULL>', source_port= 0, source_log_file='', source_log_pos= 4, source_bind=''. New state s
ource_host='<NULL>', source_port= 0, source_log_file='', source_log_pos= 4, source_bind=''.
2024-11-06T10:03:19.391449+08:00 29 [ERROR] [MY-011735] [Repl] Plugin group_replication reported: '[GCS] Unable to join the group: peers not configured. '
2024-11-06T10:03:19.391512+08:00 29 [ERROR] [MY-011639] [Repl] Plugin group_replication reported: 'Error on group communication engine start'
2024-11-06T10:03:19.391581+08:00 29 [ERROR] [MY-011735] [Repl] Plugin group_replication reported: '[GCS] The member is leaving a group without being on one.'

关键信息为:[Repl] Plugin group_replication reported: '[GCS] Unable to join the group: peers not configured.

这个错误信息意味着没有配置系统变量group_replication_group_seeds。 另外,手工启动组复制,错误日志依然报同样的错误,执行启动组复制命令的错误信息如下所示:

mysql> start group_replication;
ERROR 3097 (HY000): The START GROUP_REPLICATION command failed as there was an error when joining the communication group.
mysql> 

检查发现系统变量group_replication_group_seeds的值为空(my.cnf中没有设置这个系统变量,在mysqld-auto.cnf中系统变量为空值),因为是测试环境,不清楚什么时候动过这个系统变量,,毕竟测试环境经常被折腾来折腾去。

mysql> show variables like 'group_replication_local_address';
+---------------------------------+---------------------+
| Variable_name                   | Value               |
+---------------------------------+---------------------+
| group_replication_local_address | mysqldbu01:7306     |
+---------------------------------+---------------------+
1 row in set (0.01 sec)

mysql> 
mysql>  show variables like 'group_replication_group_seeds';
+-------------------------------+-------+
| Variable_name                 | Value |
+-------------------------------+-------+
| group_replication_group_seeds |       |
+-------------------------------+-------+
1 row in set (0.00 sec)

mysql> select * from performance_schema.persisted_variables where variable_name='group_replication_group_seeds';
+-------------------------------+----------------+
| VARIABLE_NAME                 | VARIABLE_VALUE |
+-------------------------------+----------------+
| group_replication_group_seeds |                |
+-------------------------------+----------------+
1 row in set (0.01 sec)

mysql> 

设置系统变量group_replication_group_seeds的值后,重启组复制就正常了。

mysql> set persist group_replication_group_seeds='mysqldbu02:7306';
Query OK, 0 rows affected (0.01 sec)

mysql> start group_replication;
Query OK, 0 rows affected, 1 warning (11.28 sec)

mysql> 
作者:潇湘隐者原文地址:https://www.cnblogs.com/kerrycode/p/18531222

%s 个评论

要回复文章请先登录注册