apt-get install -y percona-toolkit
# 检查重复索引
pt-duplicate-key-checker -host=’sohu.dev.chenpeng.info’ -port=’43307′ -databases=’grape’
# 运行状态监控
pt-mysql-summary -host=’sohu.dev.chenpeng.info’ -port=’43307′ -databases=’grape’
# 在线修改表结构
pt-online-schema-change -host=’sohu.dev.chenpeng.info’ -port=’3306′ -alter “ADD COLUMN c1 INT” D=chenpeng_litchi,t=lc_operate_log -execute -print
pt-online-schema-change -host=’sohu.dev.chenpeng.info’ -port=’3306′ -alter “ENGINE=InnoDB” D=chenpeng_litchi,t=lc_operate_log -execute -print
# 格式化explain结果
mysql -hsohu.dev.chenpeng.info -P43307 grape -e”explain SELECT a.line_id,b.dnf_source FROM gp_report_advertiser_day a INNER JOIN gp_line b ON a.line_id=b.id WHERE a.data_date=’2016-04-10′ LIMIT 100″ | pt-visual-explain
重复索引结果示例：
# ########################################################################
# grape.gp_mars_report_advertiser_hour
# ########################################################################
# line_id is a left-prefix of line_id_data_time
# Key definitions:
# KEY `line_id` (`line_id`) USING BTREE,
# KEY `line_id_data_time` (`line_id`,`data_time`) USING BTREE
# Column types:
# `line_id` bigint(20) unsigned not null comment ‘??id’
# `data_time` datetime not null comment ‘??’
# To remove this duplicate index, execute:
ALTER TABLE `grape`.`gp_mars_report_advertiser_hour` DROP INDEX `line_id`;
# ########################################################################
# grape.gp_operate_log
# ########################################################################
# related_id is a left-prefix of search3
# Key definitions:
# KEY `related_id` (`related_id`),
# KEY `search3` (`related_id`,`opt_datetime`),
# Column types:
# `related_id` int(10) unsigned not null comment ‘?????id’
# `opt_datetime` datetime not null comment ‘??????’
# To remove this duplicate index, execute:
ALTER TABLE `grape`.`gp_operate_log` DROP INDEX `related_id`;
# ########################################################################
# grape.gp_report_advertiser_hour
# ########################################################################
# idx_line_id is a left-prefix of line_id_data_time
# Key definitions:
# KEY `idx_line_id` (`line_id`),
# KEY `line_id_data_time` (`line_id`,`data_time`),
# Column types:
# `line_id` bigint(20) not null
# `data_time` datetime not null
# To remove this duplicate index, execute:
ALTER TABLE `grape`.`gp_report_advertiser_hour` DROP INDEX `idx_line_id`;
# ########################################################################
# Summary of indexes
# ########################################################################
# Size Duplicate Indexes 48
# Total Duplicate Indexes 3
# Total Indexes 268
工具检测到gp_operate_log有重复的索引，提供以下SQL语句删除重复的索引：
ALTER TABLE `grape`.`gp_operate_log` DROP INDEX `related_id`;
explain简化显示示例：
root@ubuntu:/tmp/d3OnxmMr# mysql -hsohu.dev.chenpeng.info -P43307 -ugrapeuser -pfaf74223tr grape -e”explain SELECT a.line_id,b.dnf_source FROM gp_report_advertiser_day a INNER JOIN gp_line b ON a.line_id=b.id WHERE a.data_date=’2016-04-10′ LIMIT 100″ | pt-visual-explain
JOIN
+- Filter with WHERE
| +- Bookmark lookup
| +- Table
| | table b
| | possible_keys PRIMARY
| +- Unique index lookup
| key b->PRIMARY
| possible_keys PRIMARY
| key_len 4
| ref grape.a.line_id
| rows 1
+- Index lookup
key a->data_date
possible_keys line_id,data_date
key_len 3
ref const
rows 10722