Troubleshooting MySQL 8 Redo Log File IO operations

Redo log file IO operations can significantly impact the performance of a MySQL database. The redo log is used to recover the database after a crash or unexpected shutdown. It records all changes made to the database, including insert, update, and delete operations. This means that any time a transaction is committed, the redo log must be flushed to disk. If the redo log is not written to disk fast enough, the database will become stalled, and performance will suffer.

If the redo log is too small, it will be written to frequently, leading to excessive disk I/O and decreased performance. On the other hand, if the redo log is too large, the database may not be able to flush it fast enough, leading to a large amount of unwritten data, and increased risk of data loss in the event of a crash.

It is essential to monitor the redo log file IO operations and adjust the size of the redo log files as needed to ensure optimal performance. Monitoring the number of read and write operations, the time spent waiting for disk I/O, and the size of the redo log files can provide valuable insights into the performance of the database and help identify potential performance bottlenecks.

Monitoring IO Operations happening on Redo Log Files 

The query copied below provides a comprehensive view of the performance of the InnoDB redo log files, including information about the file I/O operations, the events and locks related to the operations, and the current statements being executed:

select
e.SOURCE as Source,
FORMAT(e.CPU_TIME / 1000000000, 2) AS CPU_TIME_SECONDS,
irlf.FILE_NAME as File_Name,
irlf.START_LSN as Starting_LSN,
irlf.END_LSN as Enging_LSN,
fsbi.COUNT_WRITE as Total_WRITE_Operations,
fsbi.COUNT_READ as Total_READ_Operations,
fsbi.COUNT_STAR as Total_IO_Operations,
esc.EVENT_NAME as Event_Name,
dl.LOCK_TYPE as Lock_Type,
dl.LOCK_MODE as Lock_Mode,
dl.LOCK_STATUS as Lock_Status,
FROM_UNIXTIME(esc.TIMER_START, ‘%Y-%m-%d %H:%i:%s’) AS Timer_Start,
format(esc.TIMER_WAIT / 1000000000, 2) AS Timer_Wait_Sec
from
performance_schema.innodb_redo_log_files irlf
left join file_summary_by_instance fsbi on irlf.FILE_NAME = fsbi.FILE_NAME
left join events_stages_current esc on fsbi.EVENT_NAME = esc.EVENT_NAME
left join data_locks dl on esc.EVENT_ID = dl.EVENT_ID
left join events_statements_current e on esc.THREAD_ID = e.THREAD_ID;

  • Source: The source of the current statement being executed.
  • CPU_TIME_SECONDS: The CPU time used by the current statement, in seconds.
  • File_Name: The name of the InnoDB redo log file.
  • Starting_LSN: The starting LSN of the redo log file.
  • Enging_LSN: The ending LSN of the redo log file.
  • Total_WRITE_Operations: The total number of write operations performed on the file.
  • Total_READ_Operations: The total number of read operations performed on the file.
  • Total_IO_Operations: The total number of I/O operations performed on the file.
  • Event_Name: The name of the event related to the file I/O operations.
  • Lock_Type: The type of lock related to the event.
  • Lock_Mode: The mode of the lock.
  • Lock_Status: The status of the lock.
  • Timer_Start: The start time of the event.
  • Timer_Wait_Sec: The wait time for the event, in seconds.
About Shiv Iyer 422 Articles
Open Source Database Systems Engineer with a deep understanding of Optimizer Internals, Performance Engineering, Scalability and Data SRE. Shiv currently is the Founder, Investor, Board Member and CEO of multiple Database Systems Infrastructure Operations companies in the Transaction Processing Computing and ColumnStores ecosystem. He is also a frequent speaker in open source software conferences globally.