db mysql / mysql cluster 5.7.19 / my.cnf / max_binlog_cache_size / binlog

s

mysql修改binlog保存的天数
https://blog.csdn.net/Hu_wen/article/details/80582013
查看binlog过期时间,设置的时间为90天,这个值默认是0天,也就是说不自动清理,可以根据生产情况修改,本例修改为7天
mysql> show variables like 'expire_logs_days';
+------------------+-------+
| Variable_name | Value |
+------------------+-------+
| expire_logs_days | 90 |
+------------------+-------+
1 row in set (0.00 sec)

mysql> set global expire_logs_days=7;
Query OK, 0 rows affected (0.00 sec)
设置之后不会立即清除,触发条件是:

binlog大小超过max_binlog_size
手动执行flush logs
重新启动时(MySQL将会new一个新文件用于记录binlog)
我们执行flush logs;
mysql> flush logs;
Query OK, 0 rows affected, 64 warnings (0.16 sec
如果binlog非常多,不要轻易设置改参数,有可能导致io争用,这时候可以使用purge命令予以清除:

将bin.000055之前的binlog清掉:
mysql>purge binary logs to 'bin.000055';
将指定时间之前的binlog清掉:
mysql>purge binary logs before '2017-05-01 13:09:51';

问题1:

/dev/mapper/datavg-binloglv
                       20G   19G     0 100% /binlog
/dev/mapper/datavg-datalv
                       29G   22G  5.3G  81% /mysql
[root@sctssitdb18 ~]# cd /binlog/
[root@sctssitdb18 binlog]# ll
total 19418016
drwx------ 2 mysql mysql      16384 Feb  7 15:03 lost+found
-rw-r----- 1 mysql mysql 3217373871 Apr 29 11:29 mysql-bin.000005
-rw-r----- 1 mysql mysql 3150164739 Apr 29 12:21 mysql-bin.000006
-rw-r----- 1 mysql mysql 1931461984 Apr 29 12:29 mysql-bin.000007
-rw-r----- 1 mysql mysql 1783244637 Apr 29 12:30 mysql-bin.000008
-rw-r----- 1 mysql mysql 1366259526 Apr 29 12:41 mysql-bin.000009
-rw-r----- 1 mysql mysql 3108114143 May  5 17:55 mysql-bin.000010
-rw-r----- 1 mysql mysql 2151489577 May  5 18:21 mysql-bin.000011
-rw-r----- 1 mysql mysql 2682695927 May  6 09:17 mysql-bin.000012
-rw-r----- 1 mysql mysql  473747456 May  6 10:56 mysql-bin.000013
-rw-r----- 1 mysql mysql        279 May  6 09:17 mysql-bin.index

解决方案:删除非近期的mysql-bin.xxxx 文件即可,如有集群mysql最好用命令清理binlog文件,容易出现问题,万一要恢复需要用binlog。

mysql报错Multi-statement transaction required more than 'max_binlog_cache_size' bytes of storage

https://www.cnblogs.com/MYSQLZOUQI/p/6820688.html

Part1:max_binlog_cache_size = 4G ?

http://blog.51cto.com/suifu/1859252?zk1

浅析mysql内存参数之-- binlog_cache_size

https://blog.csdn.net/lxpbs8851/article/details/38455223

原my.cnf文件

内容:

55 # BINARY LOGGING # 
56 log_bin = /mysql/binlog/mysql-bin
57 binlog_format = ROW
58 expire_logs_days = 14
59 sync_binlog = 1
60 binlog_cache_size = 1M
61 max_binlog_cache_size = 2G

数据库查询

mysql> show global status like 'bin%';
+----------------------------+------------------+
| Variable_name | Value |
+----------------------------+------------------+
| Binlog_snapshot_file | mysql-bin.000232 |
| Binlog_snapshot_position | 987410213 |
| Binlog_cache_disk_use | 15588 |
| Binlog_cache_use | 19931056 |
| Binlog_stmt_cache_disk_use | 0 |
| Binlog_stmt_cache_use | 10 |
+----------------------------+------------------+

异常内容:

2018-08-04 07:00:00,001 INFO [stdout] (pool-5-thread-1) 07:00:00.001 [pool-5-thread-1] INFO c.s.p.s.service.TmpDateCleanJob - 定时JOB,清理perfStatisticsTemp#开始
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) 07:01:32.870 [pool-5-thread-1] ERROR o.s.s.s.TaskUtils$LoggingErrorHandler - Unexpected error occurred in scheduled task.
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) org.springframework.jdbc.UncategorizedSQLException: 
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) ### Error updating database. Cause: java.sql.SQLException: Multi-statement transaction required more than 'max_binlog_cache_size' bytes of storage; increase this mysqld variable and try again
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) ### The error may involve cn.s*****.ptsp.dao.mapper.PerfStatisticsMapper.deleteByPerfIds-Inline
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) ### The error occurred while setting parameters
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) ### SQL: DELETE FROM perf_statistics_temp WHERE perf_id IN ( ? )
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) ### Cause: java.sql.SQLException: Multi-statement transaction required more than 'max_binlog_cache_size' bytes of storage; increase this mysqld variable and try again
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) ; uncategorized SQLException for SQL []; SQL state [HY000]; error code [1197]; Multi-statement transaction required more than 'max_binlog_cache_size' bytes of storage; increase this mysqld variable and try again; nested exception is java.sql.SQLException: Multi-statement transaction required more than 'max_binlog_cache_size' bytes of storage; increase this mysqld variable and try again
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at org.springframework.jdbc.support.AbstractFallbackSQLExceptionTranslator.translate(AbstractFallbackSQLExceptionTranslator.java:84)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at org.springframework.jdbc.support.AbstractFallbackSQLExceptionTranslator.translate(AbstractFallbackSQLExceptionTranslator.java:81)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at org.springframework.jdbc.support.AbstractFallbackSQLExceptionTranslator.translate(AbstractFallbackSQLExceptionTranslator.java:81)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at org.mybatis.spring.MyBatisExceptionTranslator.translateExceptionIfPossible(MyBatisExceptionTranslator.java:73)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at org.mybatis.spring.SqlSessionTemplate$SqlSessionInterceptor.invoke(SqlSessionTemplate.java:368)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at com.sun.proxy.$Proxy32.delete(Unknown Source)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at org.mybatis.spring.SqlSessionTemplate.delete(SqlSessionTemplate.java:268)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at org.apache.ibatis.binding.MapperMethod.execute(MapperMethod.java:58)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at org.apache.ibatis.binding.MapperProxy.invoke(MapperProxy.java:53)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at com.sun.proxy.$Proxy34.deleteByPerfIds(Unknown Source)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at cn.s*****.ptsp.service.service.TmpDateCleanJob.perfStatisticsTempCleaner(TmpDateCleanJob.java:55)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at java.lang.reflect.Method.invoke(Method.java:498)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at org.springframework.scheduling.support.ScheduledMethodRunnable.run(ScheduledMethodRunnable.java:65)
2018-08-04 07:01:32,872 INFO [stdout] (pool-5-thread-1) at org.springframework.scheduling.support.DelegatingErrorHandlingRunnable.run(DelegatingErrorHandlingRunnable.java:54)
2018-08-04 07:01:32,873 INFO [stdout] (pool-5-thread-1) at org.springframework.scheduling.concurrent.ReschedulingRunnable.run(ReschedulingRunnable.java:81)
2018-08-04 07:01:32,873 INFO [stdout] (pool-5-thread-1) at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
2018-08-04 07:01:32,873 INFO [stdout] (pool-5-thread-1) at java.util.concurrent.FutureTask.run(FutureTask.java:266)
2018-08-04 07:01:32,873 INFO [stdout] (pool-5-thread-1) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$201(ScheduledThreadPoolExecutor.java:180)
2018-08-04 07:01:32,873 INFO [stdout] (pool-5-thread-1) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293)
2018-08-04 07:01:32,873 INFO [stdout] (pool-5-thread-1) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
2018-08-04 07:01:32,874 INFO [stdout] (pool-5-thread-1) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
2018-08-04 07:01:32,874 INFO [stdout] (pool-5-thread-1) at java.lang.Thread.run(Thread.java:745)
2018-08-04 07:01:32,874 INFO [stdout] (pool-5-thread-1) Caused by: java.sql.SQLException: Multi-statement transaction required more than 'max_binlog_cache_size' bytes of storage; increase this mysqld variable and try again
2018-08-04 07:01:32,874 INFO [stdout] (pool-5-thread-1) at com.mysql.jdbc.SQLError.createSQLException(SQLError.java:965)
2018-08-04 07:01:32,874 INFO [stdout] (pool-5-thread-1) at com.mysql.jdbc.MysqlIO.checkErrorPacket(MysqlIO.java:3973)
2018-08-04 07:01:32,874 INFO [stdout] (pool-5-thread-1) at com.mysql.jdbc.MysqlIO.checkErrorPacket(MysqlIO.java:3909)
2018-08-04 07:01:32,874 INFO [stdout] (pool-5-thread-1) at com.mysql.jdbc.MysqlIO.sendCommand(MysqlIO.java:2527)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at com.mysql.jdbc.ServerPreparedStatement.serverExecute(ServerPreparedStatement.java:1283)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at com.mysql.jdbc.ServerPreparedStatement.executeInternal(ServerPreparedStatement.java:783)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at com.mysql.jdbc.PreparedStatement.execute(PreparedStatement.java:1197)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at org.jboss.jca.adapters.jdbc.WrappedPreparedStatement.execute(WrappedPreparedStatement.java:404)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at org.apache.ibatis.executor.statement.PreparedStatementHandler.update(PreparedStatementHandler.java:45)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at org.apache.ibatis.executor.statement.RoutingStatementHandler.update(RoutingStatementHandler.java:73)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at org.apache.ibatis.executor.SimpleExecutor.doUpdate(SimpleExecutor.java:49)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at org.apache.ibatis.executor.BaseExecutor.update(BaseExecutor.java:115)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at org.apache.ibatis.executor.CachingExecutor.update(CachingExecutor.java:75)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at org.apache.ibatis.session.defaults.DefaultSqlSession.update(DefaultSqlSession.java:170)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at org.apache.ibatis.session.defaults.DefaultSqlSession.delete(DefaultSqlSession.java:185)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at java.lang.reflect.Method.invoke(Method.java:498)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) at org.mybatis.spring.SqlSessionTemplate$SqlSessionInterceptor.invoke(SqlSessionTemplate.java:358)
2018-08-04 07:01:32,884 INFO [stdout] (pool-5-thread-1) ... 20 common frames omitted

解决方案一

Caused by: java.sql.SQLException: Multi-statement transaction required more than 'max_binlog_cache_size' bytes of storage; increase this mysqld variable and try again

 解决方案二

开发拆分事务

mysql5.6之my.cnf配置详解

https://blog.csdn.net/sbpgywgi/article/details/78561212

my.cnf
[mysqld]
# GENERAL #
user                            = mysql
port                            = 3306
default_storage_engine          = InnoDB
socket                          = /tmp/mysql.sock
tmpdir                          = /dev/shm  放到共享内存里面,效率更高,  共享内存默认是物理内存的一半
character-set-server            = gbk     
collation-server                = gbk_chinese_ci  (校验级,不建议设置,让系统自己选择)
init-connection="set names gbk";
如果指定非utf8  和laint 1的话 建议添加 init-connection="set names gbk"; 这个
新创建的连接客户端要设计他的字符集
http://imysql.com/tag/字符集   参考
http://imysql.cn/2013/10/29/misunderstand-about-charset-handshake.shtml 参考
| utf8_general_ci          | utf8     |  33 | Yes     | Yes      |       1 | utf8默认的字符集编码
net_buffer_length               = 1024K         
skip-name-resolve
skip-external-locking
ft_min_word_len = 4
event_scheduler = 0
tmpdir                          = /dev/shm   指向共享内存 如果内存比较小需要注意
skip-character-set-client-handshake  忽略客户端连接过来的字符集,使用服务器端指定的字符集
## File
back_log                        = 500   比如我现在有1000个连接过来 数据库现在只能连接100个进程,另外900缓存起来存到back_log中
open_files_limit                = 65535 不超过内核的上限值就可以 ,但是不要设置太小   
# MyISAM #
key_buffer_size                 = 64M      只针对Myisam引擎以Innodb表为主 这个 就足够用了 如果以myisam表为主这个需要设置大些
#myisam_recover_options          = FORCE,BACKUP
myisam_sort_buffer_size         = 32M
myisam_repair_threads = 1
# SAFETY #
max_allowed_packet              = 128M    这个参数一般给默认值就可以了,设置导入导出环境这块设置不一样可能会报错。
skip_name_resolve
#buffers & cache
table_open_cache      = 2048    
table_definition_cache = 2048
max_heap_table_size = 96M
tmp_table_size = 96M
如果排序多的话下面2个参数加大一些
sort_buffer_size = 2M
join_buffer_size = 2M
thread_cache_size = 512   有点小可以加大一些 ,要根据我们的情况来顶,如果前端使用连接池512已经够用了,如果使用短链接需要加大 可以设置生1M
query_cache_size = 0     =0 是关闭掉
query_cache_type = 0
query_cache_limit = 256K
query_cache_min_res_unit = 512
thread_stack = 192K
read_buffer_size = 2M
read_rnd_buffer_size = 2M
bulk_insert_buffer_size = 32M
#connecter 
max_connections                = 1100    数据库最大连接数
max_user_connections           = 1000   用户最大连接数
max_connect_errors              = 1000000   输入错误多少次 锁定用户
#timeout
wait_timeout                  = 100   连接池一般是长连接,长连接 这个值就不要设置太小,太小的话超过一段时间就会自动踢掉了,可以根据
实际情况设置时间
interactive_timeout           = 100
connect_timeout              = 20   连接超时,设置太大会有问题,如果网络有问题就    
slave-net-timeout            = 20  这个可以设置小一点,太长会导致连接进程等待
transaction_isolation = READ-COMMITTED   事物隔离级别,交易类的用RR,如果代码曾有能力解决幻读用rc
#performance_schema=0
 
# BINARY LOGGING #
log_bin                         = /database/mysql/binlog/mysql-bin   日志存放位置
expire_logs_days                = 6    日志存放天数
sync_binlog                     = 0          sync_binlog减小io开销使用  主从必须设置为1 好保证主从一致性, 
设置为1 影响性能 ,设置为0 速度快数据不安全 可设置为0,1235,,10
binlog_format                   = row
max_binlog_size                 = 1024M
# LOGGING #
log_error                       = /database/mysql/logs/mysql-error.log    错误日志的位置
#log_queries_not_using_indexes  = 1
slow_query_log                 = 1
slow_query_log_file            = /database/mysql/logs/mysql-slow.log    慢日志的位置
long_query_time                = 1  慢查询的时候 可以小一点  ,每个sql 最好不要超过20毫秒
log-output                     = file     控制慢日志输出的   file|table  可以文件,也可表
log_warnings                   = 2     log_warnings= 2,可以避免Aborted connection
#thread pool
thread_handling=pool-of-threads 
thread_pool_oversubscribe=3  意思是一个cpu核心同时可以处理多少个并发线程的链接
 thread_pool_oversubscribe = cpu核心数  这个值不建议超过3
# INNODB #
#innodb_adaptive_flushing_method = estimate
innodb_adaptive_flushing        = 1  顺序io 的参数   自适应刷新,开启比较合理
innodb_data_home_dir            = /database/mysql/ibdata
innodb_log_group_home_dir       = /database/mysql/ibdata_log
innodb_data_file_path           = ibdata1:2048M:autoextend    共享表空间的大小
#innodb_adaptive_hash_index_partitions= 4
innodb_additional_mem_pool_size = 16M
innodb_flush_method             = O_DIRECT   
innodb_log_files_in_group       = 2
innodb_log_file_size            = 512M
innodb_log_buffer_size          = 16M
innodb_flush_log_at_trx_commit  = 1  innodb_flush_log_at_trx_commit = 0 非交易类,=1 交易类,保持一致性要设置成1 否则设置为0
innodb_file_per_table           = 1
innodb_buffer_pool_size         = 32G
innodb_buffer_pool_instances    = 8    比如场景为32 G 划分为4个相当于每个都有8个G子分区来进行管理,但是前提条件是内存一定要大于8G,也就是说切分完之后每个
都要大于1个G,大内存   高并发场景 设置要好点
innodb_blocking_buffer_pool_restore = ON
innodb_buffer_pool_restore_at_startup = 300
innodb_merge_sort_block_size    = 2M
innodb_kill_idle_transaction    = 300  查询或者dml语句超过一定时间内都会被kill掉 建议不设置
innodb_max_dirty_pages_pct      = 65
innodb_io_capacity              = 300  根据硬盘的iops来设置   ssd可以设置为2000普通的设置300-500
innodb_blocking_buffer_pool_restore = ON
innodb_buffer_pool_restore_at_startup = 300
innodb_merge_sort_block_size    = 2M
https://bugs.launchpad.net/percona-server/+bug/1072538
innodb_thread_concurrency       = 0  nnodb_thread_concurrency       = 0
一般是设置0或者CPU*2  现在已经基本上不用太关心了5.6开始当做透明的就好了
innodb_read_io_threads          = 12
innodb_write_io_threads         = 16
innodb_file_format              = Barracuda
innodb_flush_neighbor_pages     = 0    文件格式
innodb_change_buffering         = all
innodb_stats_on_metadata        = 0
innodb_lock_wait_timeout        = 30    等待锁的时间根据业务来判断
innodb_rollback_on_timeout
innodb_sync_spin_loops           = 100
innodb_spin_wait_delay           = 30
innodb_support_xa               = 0
innodb_concurrency_tickets      = 500   默认值500
log_bin_trust_function_creators = 1  
 log_bin_trust_function_creators=1:触发器和存储过程时启用。
注释:
开启binlog时,是否允许创建存储程序
(除非有SUPER权限,或者指定DETERMINISTIC、READS SQL DATA、NO SQL)
可动态修改,属于一个功能性的东西。
innodb_purge_threads            =1   purge线程  只能设置为1 5.7才可以有多个
innodb_purge_batch_size         = 32
innodb_old_blocks_pct           = 75
#Replication
server-id = 113306
#log-slave-updates
#skip_slave_start   看情况是否开启,如果作为中级节点必须开启否则不用开启,log-slave-updates 建议这个要开启
#read_only         = 1
slave-skip-errors=1032   跳过指定错误
skip_slave_start   = 1     开机自动启动同步,早起版本建议不要开启
innodb_max_dirty_pages_pct      = 35 这个参数看环境 ,建议设置为50以下,毕竟数据库以读为主,
如果设置过多,刷新的时候会造成瓶颈
#relay log
#relay-log = /database/mysql/binlog/slave-relay-bin
#relay-log-index = /database/mysql/binlog/slave-relay-bin.index
max-relay-log-size = 256M    
relay_log_purge = 1
#slave_parallel_workers = 8  并行的复制,多个数据库就要开启并行
my.cnf 里头,各个section什么意思
mysqld_safe
针对mysqld_safe这个进程或者程序一个配置的区间,这个区间只会被mysqld_safe读取
[mysql] 
只针对 mysql 这个cli客户端工具
[client]
 针对全部的client,例如mysqlmysqldumpmysqladmin都是
[mysqldump]
针对mysqldump专区
[mysqld]
mysqld这个服务进程的专区
relay_log_recovery = 1
如果是slave  的话一定要加上
[mysql]
prompt="\u@\h \D \R:\m:\s [\d]&get; "
pager="less -i -n -S"
no-auto-rehash
tee="/home/mysql/query.log"
记录数据库的操作,加上no-auto-rehash  不扫描MySQL日志
innodb_adaptive_flushing_lwm -- 设置redo log flush低水位线,当需要flush的redo log超过这个低水位线时,立即强制启用adaptive flushing,即便没有设置使用adaptive flushing机制
innodb_buffer_pool_size
innodb_additional_mem_pool_size    缓存数据字典的默认是8M  最好给16M或者32M
innodb_log_buffer_size
read_buffer_size             -- 顺序读缓冲,提高顺序读效率 
+read_rnd_buffer_size   -- 随机读缓冲,
+sort_buffer_size           -- 排序缓冲,提高排序效率 
+join_buffer_size           -- 表连接缓冲,提高表连接效率 
+binlog_cache_size       -- 二进制日志缓冲,提高二进制日志写入效率 
+tmp_table_size            -- 内存临时表,提高临时表存储效率 
+tmp_table_size            -- 内存临时表,提高临时表存取效率(一般用于group by查询) 
+thread_cache_size       -- 线程缓存,
+net_buffer_length       -- 线程持连接缓冲以及读取结果缓冲 
+net_buffer_length       -- 线程池连接缓冲以及读取结果缓冲 
需要关注的是最大连接数 不要设置太大
tmp_table_size  不要设置太大
innodb_buffer_pool_size                      -- InnoDB高速缓冲,行数据、索引缓冲,以及事务锁、自适应哈希等 
+innodb_additional_mem_pool_size    -- InnoDB数据字典额外内存,缓存所有表数据字典 
+innodb_log_buffer_size                      -- InnoDB REDO日志缓冲,提高REDO日志写入效率 
+key_buffer_size                                   -- MyISAM表索引高速缓冲,提高MyISAM表索引读写效率 
+query_cache_size                                -- 查询高速缓存,缓存查询结果,提高反复查询返回效率 
+key_buffer_size                                   -- MyISAM表索引高速缓冲,提高MyISAM表索引读写效率(建议少用MyISAM)
+query_cache_size                                -- 查询高速缓存,缓存查询结果,提高反复查询返回效率(建议关闭QC) 
+table_cahce                                         -- 表空间文件描述符缓存,提高数据表打开效率 
+table_definition_cache                        -- 表定义文件描述符缓存,提高数据表打开效率 
MyISAM和InnoDB buffer区别
InnoDB既缓存索引,又缓存数据
MyISAM只缓存索引,不缓存数据
max_heap_table_size 负责设置MEMORY/HEAP表最大容量,不管其他执行SQL产生的临时表,如果内存不够用,则不允许写入新的数据,MEMORY/HEAP表也不会转成磁盘表,只会告警超限后拒绝写入 
tmp_table_size 不负责限制 MEMORY/HEAP表最大容量,如果执行SQL产生临时表超过 tmp_table_size/max_heap_table_size,则会产生基于磁盘的MyISAM表 
| default_tmp_storage_engine                             | InnoDB  
tmp_table_size 不负责限制 MEMORY/HEAP表最大容量,如果执行SQL产生临时表超过 tmp_table_size/max_heap_table_size,则会产生基于磁盘的MyISAM表(如果default_tmp_storage_engine选项没有设定为InnoDB的话)
关闭查询缓存:
query_cache_type = 0
query_cache_size = 0
官方MySQL版本:
Total memory allocated 1098907648; in additional pool allocated 0
Dictionary memory allocated 37754
Buffer pool size   65535
Free buffers       65283
Database pages     252
Old database pages 0
Modified db pages  0
Buffer pool size 大小16kb
Buffer pool size   65535 * 16KB = 1G 
innodb_old_blocks_pct
旧的数据块的百分比
innodb_old_blocks_time
当buff_pool不够用的时候会根据时间讲旧的清除掉
新数据都是热数据过一段时间就放到存旧数据的里面,过段时间不用就清楚掉。
=======================================================
Percona分支版本:
Total memory allocated 858521600 (818M); in additional pool allocated 0
Total memory allocated by read views 88
Internal hash tables (constant factor + variable factor) -- 内部用途哈希表缓冲
    Adaptive hash index 13302240        (13281656 + 20584) -- 自适应哈希索引,约12.6M
    Page hash           830936 (buffer pool 0 only) -- data page哈希,约0.79MB,只缓存了buffer_pool_0这个instance,如果 innodb_buffer_pool_instances 设置大于0的话
    Dictionary cache    3382197         (3321904 + 60293) -- 数据字典缓冲,约3.2MB
    File system         820024  (812272 + 7752) -- 文件子系统缓冲,约0.78MB
    Lock system         2311256         (2311256 + 0) -- 锁子系统缓冲,约2.2MB
    Recovery system     0       (0 + 0)  -- 恢复子系统缓冲,目前是0
Dictionary memory allocated 60293
Buffer pool size        51199
Buffer pool size, bytes 838844416
Free buffers            50795 -- 空闲buffer,约793MB
Database pages          403   -- 缓冲热点数据的buffer,约6KB
Old database pages      0     -- 缓冲旧数据的buffer
Modified db pages       0     -- 缓冲最近修改数据的buffer 
叶金荣(4700963)21:18:01
默认128M(老版本默认8M,坑很多),以InnoDB表为主的话,设为物理内存50~70%
[MySQL FAQ]系列 -- 数据不算大,备份却非常慢,http://imysql.com/2009/09/18/mysql-faq-why-backup-is-so-slow.html 
adaptive index hash, 自适应哈希索引,用来管理buffer pool的哈希索引
adaptive index hash, size= innodb_buffer_pool / 64,随着buffer的频繁更新,会随之上升
   Lock system         2311256         (2311256 + 0) -- 锁子系统缓冲,约2.2MB
事件等待  当Innodb_buffer_pool_wait_free      就是事件等待 这个时候很有可能是内存不够 ,查看当系统的内存,如果有
剩余的内存就给buffer_pool多分配一些,如果剩余内存不多的情况下,建议添加内存或者迁移
| Innodb_buffer_pool_wait_free                  | 0            |
| Innodb_log_waits                              | 0            |
innodb_additional_mem_pool_size
数据字典以及内部数据结构缓存,表数量越多,相应的内存需要越大。
默认8M,通常设置为8~32M足够,一般建议设置为16M,如果确实不够用,那么会从系统中请求增加分配内存,并且错误日志中会提醒,目前至少还未发生过。
innodb_log_buffer_size
show global status 查看 Innodb_log_waits 是否大于0,是的话,就需要提高 innodb_log_buffer_size,否则维持原样。
show global stauts 查看30~60秒钟 Innodb_os_log_written 的间隔差异值,即可计算出 innodb_log_buffer_size 设置多大合适。
默认8M,一般设置为16 ~ 64M足够了。
max_heap_table_size 负责设置MEMORY/HEAP表最大容量,不管其他执行SQL产生的临时表,如果内存不够用,则不允许写入新的数据,MEMORY/HEAP表也不会转成磁盘表,只会告警超限后拒绝写入
tmp_table_size 不负责限制 MEMORY/HEAP表最大容量,如果执行SQL产生临时表超过 tmp_table_size/max_heap_table_size,则会产生基于磁盘的MyISAM表(如果default_tmp_storage_engine选项没有设定为InnoDB的话)
tmp_table_size            -- 内存临时表,提高临时表存取效率(一般用于group by查询)    ;max_heap_table_size 负责设置MEMORY/HEAP表最大容量,不管其他执行SQL产生的临时表,如果内存不够用,则不允许写入新的数据,MEMORY/HEAP表也不会转成磁盘表,只会告警超限后拒绝写入

end

原文地址:https://www.cnblogs.com/lindows/p/9428745.html