1. 現象

今天協助其他同學排查問題的時候，發現資料庫錯誤日誌文件已經有9G以上了，打開內容查看如下：

=====================================
2020-07-08 13:47:43 0x7fe3723ff700 INNODB MONITOR OUTPUT
=====================================
Per second averages calculated from the last 1 seconds
-----------------
BACKGROUND THREAD
-----------------
srv_master_thread loops: 28112548 srv_active, 0 srv_shutdown, 18948137 srv_idle
srv_master_thread log flush and writes: 47060685
----------
SEMAPHORES
----------
OS WAIT ARRAY INFO: reservation count 213360683
OS WAIT ARRAY INFO: reservation count 218012898
OS WAIT ARRAY INFO: reservation count 218624956
OS WAIT ARRAY INFO: reservation count 223392430
OS WAIT ARRAY INFO: reservation count 213358783
OS WAIT ARRAY INFO: reservation count 217996917
OS WAIT ARRAY INFO: reservation count 218627068
OS WAIT ARRAY INFO: reservation count 223399094
OS WAIT ARRAY INFO: reservation count 213372264
OS WAIT ARRAY INFO: reservation count 217974752
OS WAIT ARRAY INFO: reservation count 218606657
OS WAIT ARRAY INFO: reservation count 223387430
OS WAIT ARRAY INFO: reservation count 213382268
OS WAIT ARRAY INFO: reservation count 218029924
OS WAIT ARRAY INFO: reservation count 218619464
OS WAIT ARRAY INFO: reservation count 223399870
OS WAIT ARRAY INFO: signal count 2558329753
RW-shared spins 0, rounds 2208700138, OS waits 822920663
RW-excl spins 0, rounds 80631903713, OS waits 1603642807
RW-sx spins 1202513351, rounds 33533328545, OS waits 959708531
Spin rounds per wait: 2208700138.00 RW-shared, 80631903713.00 RW-excl, 27.89 RW-sx
------------------------
LATEST DETECTED DEADLOCK
------------------------
2020-04-21 19:50:05 0x7fe28a7fd700
...
...
...
Process ID=54642, Main thread ID=140614440048384, state: sleeping
Number of rows inserted 5475421722, updated 433989820, deleted 4122238559, read 669572614313
708.29 inserts/s, 34.97 updates/s, 573.43 deletes/s, 29898.10 reads/s
----------------------------
END OF INNODB MONITOR OUTPUT
============================

此內容是InnoDB MONITOR的結果（和show engine innodb status結果一樣），也就是說開啟了InnoDB監視器。

主要涉及的參數有innodb_status_output 和 innodb_status_output_locks ，這兩個系統變數是用來啟用標準innodb監控和 InnoDB鎖監控的，查看本實例參數情況：

mysql> show  global  variables like '%innodb_status%';
+----------------------------+-------+
| Variable_name              | Value |
+----------------------------+-------+
| innodb_status_output       | ON   |
| innodb_status_output_locks | ON   |
+----------------------------+-------+
2 rows in set (0.01 sec)

 說明確實已經開啟了。

2. 關閉INNODB MONITOR

可以線上關閉INNODB MONITOR，但是在關閉前建議先將原日誌備份（重命名）

mv  mysqld.log   mysqld.log.20200708

再修改參數，關閉監控

mysql> set global innodb_status_output='OFF';
Query OK, 0 rows affected (0.00 sec)

mysql> set global innodb_status_output_locks='OFF';
Query OK, 0 rows affected (0.00 sec)

mysql> flush logs;

時間有限，今天暫時就簡單說明現象及引起日誌暴漲的原因，後續有機會在關註日誌內容及相關參數意義。

想瞭解更多內容或參與技術交流可以關註微信公眾號【資料庫乾貨鋪】或進技術交流群溝通。