错误日志
2018-03-29T15:35:48.608513Z 8473251 [Note] Multi-threaded slave: Coordinator has waited 31621 times hitting slave_pending_jobs_size_max; current event size = 7990.
2018-03-29T15:35:48.618715Z 8473251 [Note] Multi-threaded slave: Coordinator has waited 31631 times hitting slave_pending_jobs_size_max; current event size = 8002.
2018-03-29T15:35:48.625974Z 8473251 [Note] Multi-threaded slave: Coordinator has waited 31641 times hitting slave_pending_jobs_size_max; current event size = 8078.
2018-03-29T15:35:48.632782Z 8473251 [Note] Multi-threaded slave: Coordinator has waited 31651 times hitting slave_pending_jobs_size_max; current event size = 7990.
2018-03-29T15:35:49.921167Z 8473252 [ERROR] Slave SQL for channel '': Worker 1 failed executing transaction 'ANONYMOUS' at master log mysql-bin.000096, end_log_pos 912768886; Could not execute Update_rows event on table zyd1.zyd1_action_log; Can't find record in 'zyd1_action_log', Error_code: 1032; handler error HA_ERR_KEY_NOT_FOUND; the event's master log mysql-bin.000096, end_log_pos 912768886, Error_code: 1032
2018-03-29T15:35:55.416105Z 8473251 [ERROR] Error writing relay log configuration.
2018-03-29T15:35:56.941455Z 8473251 [Warning] Slave SQL for channel '': ... The slave coordinator and worker threads are stopped, possibly leaving data in inconsistent state. A restart should restore consistency automatically, although using non-transactional storage for data or info tables or DDL queries could lead to problems. In such cases you have to examine your data (see documentation for details). Error_code: 1756
2018-03-29T15:35:56.941479Z 8473251 [Note] Error reading relay log event for channel '': slave SQL thread was killed
机器翻译
2018-03-29T15：35：48.608513Z 8473251 [注]多线程从机：协调器等待31621次，达到slave_pending_jobs_size_max;当前事件大小= 7990。
2018-03-29T15：35：48.618715Z 8473251 [注]多线程从机：协调器等待31631次，达到slave_pending_jobs_size_max;当前事件大小= 8002。
2018-03-29T15：35：48.625974Z 8473251 [注]多线程从机：协调器等待命中slave_pending_jobs_size_max 31641次;当前事件大小= 8078。
2018-03-29T15：35：48.632782Z 8473251 [注]多线程从机：协调器等待命中slave_pending_jobs_size_max 31651次;当前事件大小= 7990。
2018-03-29T15：35：49.921167Z 8473252 [错误]通道“从属SQL”：工人1在主日志上执行事务'ANONYMOUS'失败mysql-bin.000096，end_log_pos 912768886;无法在表zyd1.zyd1_action_log上执行Update_rows事件; 'zyd1_action_log'中找不到记录，Error_code：1032;处理程序错误HA_ERR_KEY_NOT_FOUND;事件的主日志mysql-bin.000096，end_log_pos 912768886，Error_code：1032
2018-03-29T15：35：55.416105Z 8473251 [错误]写入中继日志配置时出错。
2018-03-29T15：35：56.941455Z 8473251 [警告]通道“从属SQL”：...从属协调器和工作线程停止，可能使数据处于不一致的状态。尽管对数据或信息表或DDL查询使用非事务性存储可能会导致问题，但重新启动应自动恢复一致性。在这种情况下，您必须检查您的数据（有关详细信息，请参阅文档）。 Error_code：1756
2018-03-29T15：35：56.941479Z 8473251 [注意]错误读取通道''的中继日志事件：从属SQL线程被终止
错误分析
协调器等待31621次，达到slave_pending_jobs_size_max 可能是达到这个参数峰值,
默认值为16777216（16MB），但是slave接收到的slave_pending_jobs_size_max可能超过了,我得环境是,大批量导入,导致主从挂掉
这个参数意味着什么
在多线程复制时，在队列中Pending的事件所占用的最大内存，默认为16M，如果内存富余，或者延迟较大时，可以适当调大;
注意这个值要比主库的max_allowed_packet大
官方解释
https://dev.mysql.com/doc/refman/5.7/en/replication-options-slave.html
查一下slave_pending_jobs_size_max；
mysql> show global variables like '%slave_pending_jobs_size_max%';
+-----------------------------+----------+
| Variable_name | Value |
+-----------------------------+----------+
| slave_pending_jobs_size_max | 16777216 |
+-----------------------------+----------+
1 row in set (0.00 sec)
参数调大
# 主库锁库
stop slave；
set global slave_pending_jobs_size_max=20000000;
start slave；
再次检验
mysql> show global variables like '%slave_pending_jobs_size_max%';
+-----------------------------+----------+
| Variable_name | Value |
+-----------------------------+----------+
| slave_pending_jobs_size_max | 19999744 |
+-----------------------------+----------+
1 row in set (0.00 sec)