MySQLTroubleShoting: no operation, full disk I/O Load

(I) Environment: OS: RHEL-5.8Server: MySQL5.5Engine: InnoDB (ii) Problem: without any operation, the disk I/O load is almost full, and then 3 or 4 hours of Disk Load is still iotop part of the output: TotalDISKREAD: 32.55Ms | TotalDISKWRITE: 0.00 BsTIDPRIOUSERDISKREADD

(I) Environment: OS: RHEL-5.8 Server: MySQL 5.5 Engine: InnoDB (ii) Problem: without any operation, the disk I/O load is almost full, and after 3 or 4 hours, the DISK load is still iotop part of the output: Total DISK READ: 32.55 M/s | Total disk write: 0.00 B/sTID PRIO USER DISK READ D

(I) Environment:


OS: RHEL-5.8
Server: MySQL 5.5
Engine: InnoDB


(Ii) Problem:


Without any operation, the disk I/O load is almost full, and the disk load is still running for 3 or 4 hours.

Iotop partial output:

Total DISK READ: 32.55 M/s | Total DISK WRITE: 0.00 B/sTID  PRIO  USER     DISK READ  DISK WRITE  SWAPIN     IO>    COMMAND535  be/4 mysql       3.26 M/s    0.00 B/s  0.00 % 80.88 % mysqld --basedir=...536 be/4 mysql      29.26 M/s    0.00 B/s  0.00 % 71.93 % mysqld --basedir=.........       

(Iii) symptom:


Perform a full update Performance Test on sysbench and shut down the database after the test is completed.
However, after restarting, we can find this problem and wait for 3 or 4 hours before the problem persists.



(Iv) cause:

Innodb_fast_shutdown is set to 1.

① Innodb_fast_shutdown ==> 1: When the switch buffer is disabled, merge and purge operations are not performed, but innodb_fast_shutdown is performed in the next boot cycle: the merge change buffer and purge operations are performed when the database is closed.