pt-query-digest
可以用来分析binlog、General log、slowlog,也可以通过show processlist或者通过tcpdump抓取的MySQL协议数据来进行分析
用法:
(1)直接分析slow日志:
pt-query-digest INDEX01184W-slow.log > slow_report1.log
(2)分析最近12小时内的查询:
pt-query-digest --since=12h INDEX01184W-slow.log > slow_report2.log
(3)分析指定时间范围内的查询:
#pt-query-digest --since '2017-02-01 09:30:00' --until '2017-03-01 10:00:00' INDEX01184W-slow.log >> slow_report3.log
(4)分析指含有select语句的慢查询
pt-query-digest --filter '$event->{fingerprint} =~ m/^select/i' INDEX01184W-slow.log> slow_report4.log
(5) 针对某个用户的慢查询
pt-query-digest --filter '($event->{user} || "") =~ m/^root/i' INDEX01184W-slow.log> slow_report5.log
(6) 查询所有所有的全表扫描或full join的慢查询
pt-query-digest --filter '(($event->{Full_scan} || "") eq "yes") ||(($event->{Full_join} || "") eq "yes")' INDEX01184W-slow.log> slow_report6.log
(7)把查询保存到query_review表(#默认会创建percona_schema库和query_review表 )
pt-query-digest --user=root --password=147258 --review h=localhost INDEX01184W-slow.log
(8)把查询保存到query_history表
pt-query-digest --user=root –password=abc123 --review h=localhost INDEX01184W-slow.log
(9)通过tcpdump抓取mysql的tcp协议数据,然后再分析
tcpdump -s 65535 -x -nn -q -tttt -i any -c 1000 port 3306 > mysql.tcp.txt
pt-query-digest --type tcpdump mysql.tcp.txt> slow_report9.log
(10)分析binlog
mysqlbinlog mysql-bin.000003 > mysql-bin000003.sql
pt-query-digest --type=binlog mysql-bin000003.sql > slow_report10.log
(11)分析general log
pt-query-digest --type=genlog general.log > slow_report11.log
例:
#查询两条慢SQL:
root@localhost [(none)]>select sleep(3);
+----------+
| sleep(3) |
+----------+
| 0 |
+----------+
1 row in set (3.70 sec)
root@localhost [(none)]>select sleep(4);
+----------+
| sleep(4) |
+----------+
| 0 |
+----------+
1 row in set (4.02 sec)
root@localhost [(none)]>select sleep(8);
+----------+
| sleep(8) |
+----------+
| 0 |
+----------+
1 row in set (8.07 sec)
#查看slow日志,可以发现会记录上面两条SQL:
[root@Darren1 data]# cat slow.log
# Time: 2017-06-02T05:06:04.452125Z
# User@Host: root[root] @ localhost [] Id: 5565
# Query_time: 3.665139 Lock_time: 0.000000 Rows_sent: 1 Rows_examined: 0
SET timestamp=1496379964;
select sleep(3);
# Time: 2017-06-02T05:35:42.145231Z
# User@Host: root[root] @ localhost [] Id: 6454
# Query_time: 4.013508 Lock_time: 0.000000 Rows_sent: 1 Rows_examined: 0
SET timestamp=1496381742;
select sleep(4);
# Time: 2017-06-02T07:29:33.820712Z
# User@Host: root[root] @ localhost [] Id: 9867
# Query_time: 8.032160 Lock_time: 0.000000 Rows_sent: 1 Rows_examined: 0
SET timestamp=1496388573;
select sleep(8);
#使用pt-query-digest分析slow日志文件:
[root@Darren1 data]# pt-query-digest slow.log
# 170ms user time, 70ms system time, 24.36M rss, 204.71M vsz
# Current date: Fri Jun 2 15:30:17 2017
# Hostname: Darren1
# Files: slow.log
# Overall: 3 total, 1 unique, 0.00 QPS, 0.00x concurrency ________________
# Time range: 2017-06-02T05:06:04 to 2017-06-02T07:29:33
# Attribute total min max avg 95% stddev median
# ============ ======= ======= ======= ======= ======= ======= =======
# Exec time 16s 4s 8s 5s 8s 2s 4s
# Lock time 0 0 0 0 0 0 0
# Rows sent 3 1 1 1 1 0 1
# Rows examine 0 0 0 0 0 0 0
# Query size 45 15 15 15 15 0 15
第一部分:
Overall: 总共有多少条查询,上例为总共3个查询
unique: 对SQL进行分类,总的SQL种类,上例为1种
Time range: 查询执行的时间范围
total: 总计 min:最小 max: 最大 avg:平均
95%: 把所有值从小到大排列,位置位于95%的那个数,这个数一般最具有参考价值。
median: 中位数,把所有值从小到大排列,位置位于中间那个数。
# Profile
# Rank Query ID Response time Calls R/Call V/M Item
# ==== ================== ============== ===== ====== ===== ======
# 1 0xF9A57DD5A41825CA 15.7108 100.0% 3 5.2369 0.68 SELECT
第二部分:
对SQL进行分组,然后对各类查询的执行情况进行分析,结果按总执行时长,从大到小排序。
Response: 总的响应时间。
time: 该查询在本次分析中总的时间占比。
calls: 执行次数,即本次分析总共有多少条这种类型的查询语句。
R/Call: 平均每次执行的响应时间。
Item : 查询对象
# Query 1: 0.00 QPS, 0.00x concurrency, ID 0xF9A57DD5A41825CA at byte 409
# This item is included in the report because it matches --limit.
# Scores: V/M = 0.68
# Time range: 2017-06-02T05:06:04 to 2017-06-02T07:29:33
# Attribute pct total min max avg 95% stddev median
# ============ === ======= ======= ======= ======= ======= ======= =======
# Count 100 3
# Exec time 100 16s 4s 8s 5s 8s 2s 4s
# Lock time 0 0 0 0 0 0 0 0
# Rows sent 100 3 1 1 1 1 0 1
# Rows examine 0 0 0 0 0 0 0 0
# Query size 100 45 15 15 15 15 0 15
# String:
# Hosts localhost
# Users root
# Query_time distribution
# 1us
# 10us
# 100us
# 1ms
# 10ms
# 100ms
# 1s ################################################################
# 10s+
# EXPLAIN /*!50100 PARTITIONS*/
select sleep(8)\G
第三部分:
Databases: 库名
Users: 各个用户执行的次数(占比)
Query_time distribution : 查询时间分布图, 长短体现区间占比,本例中SQL处于1s-10s。
Tables: 查询中涉及到的表
Explain: 示例
#把分析结果记录到表中DSN
[root@Darren1 data]# pt-query-digest --user=root --password=147258 --review h=localhost slow.log
root@localhost [percona_schema]>select * from percona_schema.query_review\G
......
*************************** 2. row ***************************
checksum: 17988922643135866314
fingerprint: select sleep(?)
sample: select sleep(8)
first_seen: 2017-06-02 05:06:04
last_seen: 2017-06-02 07:29:33
reviewed_by: NULL
reviewed_on: NULL
comments: NULL
2 rows in set (0.00 sec)