1. Based on MySQL master-slave environment
Can refer https://www.cnblogs.com/tyk3201/p/12054586.html
Upload and install rpm package
|
1
2
|
cd
/usr/local/src
yum -y
install
percona-toolkit-3.1.0-2.el7.x86_64.rpm
|
Create a table in a sbtest (there is the premise of stone) database table called heartbeat
|
1
|
pt-heartbeat -S
/var/lib/mysql/mysql
.sock --database sbtest --update --create-table --daemonize
|
Download pressure measurement tool
|
1
2
|
yum -y
install
epel*
yum -y kainstall sysbench
|
Pressure began to implement experimental measurement of sbtest database
|
1
2
3
4
5
6
7
8
9
10
11
12
13
|
sysbench
/usr/share/sysbench/oltp_read_write
.lua \
> --mysql-host=
'192.168.18.140'
\
//
主的数据库的IP地址
> --mysql-user=sbtest \
//
用户
> --mysql-password=123456 \
//
用户密码
> --mysql-port=3306 \
//
主数据库的端口
> --db-driver=mysql \
//
数据库的类型
> --mysql-db=sbtest \
//
要进行压测的数据库
> --tables=10 \
//
要创建10张表
> --table-size=500000 \
//
每一张数据表中的插入50万条数据
> --report-interval=10 \
//
报告间隔
> --threads=128 \
//
开启128个线程
> --
time
=120 \
//
执行两分钟
> prepare
//
开始压测测试
|
Pressure began to really test
|
1
2
3
4
5
6
7
8
9
10
11
12
13
|
sysbench
/usr/share/sysbench/oltp_read_write
.lua \
> --mysql-host=
'192.168.18.140'
\
//
主的数据库的IP地址
> --mysql-user=sbtest \
//
用户
> --mysql-password=123456 \
//
用户密码
> --mysql-port=3306 \
//
主数据库的端口
> --db-driver=mysql \
//
数据库的类型
> --mysql-db=sbtest \
//
要进行压测的数据库
> --tables=10 \
//
要创建10张表
> --table-size=500000 \
//
每一张数据表中的插入50万条数据
> --report-interval=10 \
//
报告间隔
> --threads=128 \
//
开启128个线程
> --
time
=120 \
//
执行两分钟
> run
//
开始压测实战
|
Performed on the virtual discipline from the beginning really in pressure measured when the next command to the virtual machine and the main database simultaneously executed
|
1
|
pt-heartbeat -S
/var/lib/mysql/mysql
.sock --master-server-
id
=1 --monitor --database sbtest
|
Garbage generated data recovery
|
1
2
3
4
5
6
7
8
9
10
11
12
13
|
sysbench
/usr/share/sysbench/oltp_read_write
.lua \
> --mysql-host=
'192.168.18.140'
\
//
主的数据库的IP地址
> --mysql-user=sbtest \
//
用户
> --mysql-password=123456 \
//
用户密码
> --mysql-port=3306 \
//
主数据库的端口
> --db-driver=mysql \
//
数据库的类型
> --mysql-db=sbtest \
//
要进行压测的数据库
> --tables=10 \
//
要创建10张表
> --table-size=500000 \
//
每一张数据表中的插入50万条数据
> --report-interval=10 \
//
报告间隔
> --threads=128 \
//
开启128个线程
> --
time
=120 \
//
执行两分钟
> cleanup
//
回收垃圾信息
|
We are just running in the background
|
1
2
|
ps
aux |
grep
perl
pt-heartbeat
/usr/bin/pt-heartbeat
-S
/var/lib/mysql/mysql
.sock --database sbtest --update --create-table --stop
//
关掉进程
|
test database in the master database to create a table named z1 of
|
1
2
|
use
test
;
create table z1 (
id
int not null, name varchar(32), primary key(
id
));
|
Data from the
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
|
use
test
;
MariaDB [
test
]>
select
* from z1;
Empty
set
(0.00 sec)
MariaDB [
test
]>
set
sql_log_bin=0;
Query OK, 0 rows affected (0.00 sec)
MariaDB [
test
]>
set
global read_only=off;
Query OK, 0 rows affected (0.01 sec)
MariaDB [
test
]> insert into z1(
id
,name) values(3,
"python"
);
Query OK, 1 row affected (0.00 sec)
MariaDB [
test
]>
select
* from z1;
+----+--------+
|
id
| name |
+----+--------+
| 3 | python |
+----+--------+
1 row
in
set
(0.00 sec)
|
Back insert data on the primary
|
1
2
|
insert into z1(
id
,name) values(2,
"mysql"
);
insert into z1(
id
,name) values(3,
"python"
);
|
View from the slave back to the state there have been mistakes
|
1
2
3
4
5
6
7
8
9
10
|
show slave status \G;
有一个类似这样的错误
Last_Errno: 1062
Last_Error: Error
'Duplicate entry '
3
' for key '
PRIMARY
''
on query. Default database:
'test'
. Query:
'insert into z1(id,name) values(3,"python")'
在执行一条命令
pt-slave-restart --error-numbers=1062
//
跳过错误
再去查看一下slave的状态 就可以看到双
yes
|
Detecting data consistency from the command master
On the primary database
|
1
2
3
4
5
6
7
|
pt-table-checksum --nocheck-replication-filters --no-check-binlog-
format
--replicate=
test
.checksums --create-replicate-table --databases=
test
--tables=z1 h=192.168.18.140,u=sbtest,p=123456,P=3306
//
执行效果 说的是检测到一条没有同步
Checking
if
all tables can be checksummed ...
Starting checksum ...
TS ERRORS DIFFS ROWS DIFF_ROWS CHUNKS SKIPPED TIME TABLE
12-17T11:47:14 0 0 2 0 1 0 0.011
test
.z1
|
Never repair the master synchronization
|
1
2
3
|
pt-table-
sync
--replicate=
test
.checksums h=172.18.47.110,u=huazai007,p=123456,P=3306 h=172.18.47.112,u=huazai007,p=123456,P=3306 --print 只是打印
pt-table-
sync
--replicate=
test
.checksums h=172.18.47.110,u=huazai007,p=123456,P=3306 h=172.18.47.112,u=huazai007,p=123456,P=3306 --execute
//
校验
|