Yum将Percona从5.6.25-rel73.1更新到5.6.26-rel74.0导致服务器无法启动

Question

Running under CentOS 6.7 yum updating Percona server leaves it in an unstartable state.

This is from the log /var/log/mysqld.log

150929 14:54:17 mysqld_safe Starting mysqld daemon with databases from /srv/mysql
2015-09-29 14:54:18 0 [Warning] TIMESTAMP with implicit DEFAULT value is deprecated. Please use --explicit_defaults_for_timestamp server option (see documentation for more details).
2015-09-29 14:54:18 0 [Note] /usr/sbin/mysqld (mysqld 5.6.26-74.0) starting as process 17130 ...
2015-09-29 14:54:18 17130 [Note] Plugin 'FEDERATED' is disabled.
2015-09-29 14:54:18 17130 [Note] InnoDB: Using atomics to ref count buffer pool pages
2015-09-29 14:54:18 17130 [Note] InnoDB: The InnoDB memory heap is disabled
2015-09-29 14:54:18 17130 [Note] InnoDB: Mutexes and rw_locks use GCC atomic builtins
2015-09-29 14:54:18 17130 [Note] InnoDB: Memory barrier is not used
2015-09-29 14:54:18 17130 [Note] InnoDB: Compressed tables use zlib 1.2.3
2015-09-29 14:54:18 17130 [Note] InnoDB: Using Linux native AIO
2015-09-29 14:54:18 17130 [Note] InnoDB: Using CPU crc32 instructions
2015-09-29 14:54:18 17130 [Note] InnoDB: Initializing buffer pool, size = 3.0G
2015-09-29 14:54:18 17130 [Note] InnoDB: Completed initialization of buffer pool
2015-09-29 14:54:18 17130 [Note] InnoDB: Highest supported file format is Barracuda.
2015-09-29 14:54:18 17130 [ERROR] InnoDB: Tried to read 4096 bytes at offset 286720. Was only able to read 0.
2015-09-29 14:54:18 7fe0f5f847e0  InnoDB: Operating system error number 22 in a file operation.
InnoDB: Error number 22 means 'Invalid argument'.
InnoDB: Some operating system error numbers are described at
InnoDB: http://dev.mysql.com/doc/refman/5.6/en/operating-system-error-codes.html
2015-09-29 14:54:18 17130 [ERROR] InnoDB: File (unknown): 'read' returned OS error 122. Cannot continue operation
150929 14:54:18 mysqld_safe mysqld from pid file /var/run/mysqld/mysqld.pid ended

The my.cnf looks something like this:

[mysqld]
socket=/var/lib/mysql/mysql.sock
user=mysql
# Disabling symbolic-links is recommended to prevent assorted security risks
symbolic-links=0
innodb_track_changed_pages = true
pid-file=/var/run/mysqld/mysqld.pid

# use interleaved id creation during bulk inserts
# w/o this bulk inserts are serialized
innodb_autoinc_lock_mode=2

innodb_log_file_size = 2000M
innodb_flush_method = O_DIRECT
#innodb_buffer_pool_size = 80G
innodb_buffer_pool_size = 5G
transaction-isolation=READ-COMMITTED

innodb_file_per_table=1
default-storage-engine = InnoDB
innodb_file_format = Barracuda

The culprit seems to be this line:

innodb_track_changed_pages = true

Answer 1

In order to work around this...

1. We changed innodb_track_changed_pages to false
2. Stop the server
3. Manually delete the ib_modified_log* files in the mysql directory
4. Start the server
5. Change innodb_track_changed_pages to true
6. Restart the server

This seems like a bug.