使用HAProxy给MySQL slave群进行负载均衡和状态监控
一.安装haproxy haproxy机器 http://haproxy.1wt.deu 需* tar zxvf haproxy-1.4.25.tar.gzcd haproxy-1.4.25make TARGET=linux26make installmkdir -p /usr/local/haproxy/chown nobody:nobody /usr/local/haproxy/mkdir /etc/haproxy/cp examples/haprox
一.安装haproxy
haproxy机器
http://haproxy.1wt.deu
需*
tar zxvf haproxy-1.4.25.tar.gz cd haproxy-1.4.25 make TARGET=linux26 make install mkdir -p /usr/local/haproxy/ chown nobody:nobody /usr/local/haproxy/ mkdir /etc/haproxy/ cp examples/haproxy.cfg /etc/haproxy/ cp examples/haproxy.init /etc/init.d/haproxy chown root:root /etc/init.d/haproxy chmod 700 /etc/init.d/haproxy
修改haproxy启动脚本
/usr/sbin/$BASENAME
改成
/usr/local/sbin/$BASENAME
sed -i -r 's|/usr/sbin|/usr/local/sbin|' /etc/init.d/haproxy
编辑配置文件
vi /etc/haproxy/haproxy.cfg
global #log 127.0.0.1 local0 log 127.0.0.1 local3 info #log loghost local0 info maxconn 4096 chroot /usr/local/haproxy uid nobody gid nobody daemon debug #quiet defaults log global mode tcp #option httplog option dontlognull retries 3 option redispatch maxconn 2000 contimeout 5000 clitimeout 50000 srvtimeout 50000 frontend mysql bind 192.168.0.107:3306 maxconn 3000 default_backend mysql_slave backend mysql_slave #cookie SERVERID rewrite mode tcp balance roundrobin #balance source #balance leastconn contimeout 10s timeout check 2s option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www server mysql_192_168_0_104_3306 192.168.0.104:3306 weight 1 check port 9300 inter 5s rise 2 fall 3 server mysql_192_168_0_104_3307 192.168.0.104:3307 weight 1 check port 9301 inter 5s rise 2 fall 3 #server mysql_192_168_0_106_3306 192.168.0.106:3306 weight 1 check port 9300 inter 5s rise 2 fall 3 listen admin_status mode http bind 192.168.0.107:8000 option httplog log global stats enable stats refresh 30s stats hide-version stats realm Haproxy\ Statistics stats uri /admin-status stats auth admin:123456 stats admin if TRUE
打开监控的iptables
iptables -A INPUT -p tcp -m tcp -s 192.168.0.0/24 --dport 8000 -j ACCEPT
添加自启动并启动haproxy服务
chkconfig –add haproxy chkconfig haproxy on service haproxy start
被监控机上
我这里是单机双实例,所以有2个脚本,单机只需一个脚本和一个服务端口就行
编辑mysql检测3306脚本
vi /opt/shell/mysqlchk_status_3306.sh
#!/bin/bash
#
# /usr/local/bin/mysqlchk_status.sh
#
# This script checks if a mysql server is healthy running on localhost. It will
# return:
#
# "HTTP/1.x 200 OK\r" (if mysql is running smoothly)
#
# – OR –
#
# "HTTP/1.x 503 Internal Server Error\r" (else)
#
MYSQL_HOST="localhost"
MYSQL_PORT="3306"
MYSQL_USERNAME="mysqlcheck"
MYSQL_PASSWORD="paSSword"
MYSQL_PATH="/opt/mysql/bin/"
#
# We perform a simple query that should return a few results
#${MYSQL_PATH}mysql -h${MYSQL_HOST} -P${MYSQL_PORT} -u${MYSQL_USERNAME} -p${MYSQL_PASSWORD} -e "show slave status\G;" >/tmp/rep${MYSQL_PORT}.txt
${MYSQL_PATH}mysql -h${MYSQL_HOST} -P${MYSQL_PORT} -u${MYSQL_USERNAME} -p${MYSQL_PASSWORD} -e "show full processlist;" >/tmp/processlist${MYSQL_PORT}.txt
${MYSQL_PATH}mysql -h${MYSQL_HOST} -P${MYSQL_PORT} -u${MYSQL_USERNAME} -p${MYSQL_PASSWORD} -e "show slave status\G;" >/tmp/rep${MYSQL_PORT}.txt
iostat=`grep "Slave_IO_Running" /tmp/rep${MYSQL_PORT}.txt |awk '{print $2}'`
sqlstat=`grep "Slave_SQL_Running" /tmp/rep${MYSQL_PORT}.txt |awk '{print $2}'`
result=$(cat /tmp/processlist${MYSQL_PORT}.txt|wc -l)
echo iostat:$iostat and sqlstat:$sqlstat
# if slave_IO_Running and Slave_sql_Running ok,then return 200 code
if [ "$result" -gt "3" ] && [ "$iostat" = "Yes" ] && [ "$sqlstat" = "Yes" ];
then
# mysql is fine, return http 200
/bin/echo -e "HTTP/1.1 200 OK\r\n"
else
# mysql is down, return http 503
/bin/echo -e "HTTP/1.1 503 Service Unavailable\r\n"
fi
vi /opt/shell/mysqlchk_status_3307.sh
#!/bin/bash
#
# /usr/local/bin/mysqlchk_status.sh
#
# This script checks if a mysql server is healthy running on localhost. It will
# return:
#
# "HTTP/1.x 200 OK\r" (if mysql is running smoothly)
#
# – OR –
#
# "HTTP/1.x 503 Internal Server Error\r" (else)
#
MYSQL_HOST="localhost"
MYSQL_PORT="3307"
MYSQL_USERNAME="mysqlcheck"
MYSQL_PASSWORD="paSSword"
MYSQL_PATH="/opt/mysql/bin/"
#
# We perform a simple query that should return a few results
#${MYSQL_PATH}mysql -h${MYSQL_HOST} -P${MYSQL_PORT} -u${MYSQL_USERNAME} -p${MYSQL_PASSWORD} -e "show slave status\G;" >/tmp/rep${MYSQL_PORT}.txt
${MYSQL_PATH}mysql -S/data/mysql/mysql.sock -u${MYSQL_USERNAME} -p${MYSQL_PASSWORD} -e "show full processlist;" >/tmp/processlist${MYSQL_PORT}.txt
${MYSQL_PATH}mysql -S/data/mysql/mysql.sock -u${MYSQL_USERNAME} -p${MYSQL_PASSWORD} -e "show slave status\G;" >/tmp/rep${MYSQL_PORT}.txt
iostat=`grep "Slave_IO_Running" /tmp/rep${MYSQL_PORT}.txt |awk '{print $2}'`
sqlstat=`grep "Slave_SQL_Running" /tmp/rep${MYSQL_PORT}.txt |awk '{print $2}'`
result=$(cat /tmp/processlist${MYSQL_PORT}.txt|wc -l)
#echo iostat:$iostat and sqlstat:$sqlstat
echo $result
# if slave_IO_Running and Slave_sql_Running ok,then return 200 code
if [ "$result" -gt "3" ] && [ "$iostat" = "Yes" ] && [ "$sqlstat" = "Yes" ];
then
# mysql is fine, return http 200
/bin/echo -e "HTTP/1.1 200 OK\r\n"
else
# mysql is down, return http 503
/bin/echo -e "HTTP/1.1 503 Service Unavailable\r\n"
fi
chmod 775 /opt/shell/mysqlchk_status_3306.sh
chmod 775 /opt/shell/mysqlchk_status_3307.sh
在mysql slave另行建立一个具有process和slave_client权限的账号。
CREATE USER 'mysqlcheck'@'localhost' IDENTIFIED BY 'PaSSword'; GRANT PROCESS , REPLICATION CLIENT ON * . * TO 'mysqlcheck'@'localhost' IDENTIFIED BY 'PaSSword' WITH MAX_QUERIES_PER_HOUR 0 MAX_CONNECTIONS_PER_HOUR 0 MAX_UPDATES_PER_HOUR 0 MAX_USER_CONNECTIONS 0 ; flush privileges;
测试脚本
./mysqlchk_status_3306.sh
添加服务
绑定内网ip,运行于930端口,只开放给192.168.0内网
yum install -y xinetd
vim /etc/xinetd.d/mysql_status
service mysqlchk_status3306
{
flags = REUSE
socket_type = stream
bind = 192.168.0.104
port = 9300
wait = no
user = nobody
server = /opt/shell/mysqlchk_status_3306.sh
log_type = FILE /dev/null
log_on_failure += USERID
disable = no
only_from = 192.168.0.0/24
}
service mysqlchk_status3307
{
flags = REUSE
socket_type = stream
bind = 192.168.0.104
port = 9301
wait = no
user = nobody
server = /opt/shell/mysqlchk_status_3307.sh
log_type = FILE /dev/null
log_on_failure += USERID
disable = no
only_from = 192.168.0.0/24
}
bind和only_from的ip地址要有haproxy能请求的权限,使用drbd用0.0.0.0
user要用server脚本的执行权限
port端口要在/etc/service 中声明
chattr -i /etc/services
vi /etc/services
mysqlchk_status3306 9300/tcp #haproxy mysql check mysqlchk_status3307 9301/tcp #haproxy mysql check
services中的mysqlchk_status3306 要和xinetd.d中service名对应
打开iptables
iptables -A INPUT -p tcp -m tcp -s 192.168.0.0/24 --dport 9300 -j ACCEPT iptables -A INPUT -p tcp -m tcp -s 192.168.0.0/24 --dport 9301 -j ACCEPT
/etc/init.d/iptables save
添加自启动及启动服务
chkconfig xinetd –level 345 on
/etc/init.d/xinetd start
查看是否运行
netstat -lntp
Active Internet connections (only servers) Proto Recv-Q Send-Q Local Address Foreign Address State PID/Program name tcp 0 0 0.0.0.0:9300 0.0.0.0:* LISTEN 4863/xinetd tcp 0 0 0.0.0.0:9301 0.0.0.0:* LISTEN 4863/xinetd
如果没有的话注意检测下bind地址及服务端口
在监控机运行测试
telnet 192.168.0.104 9300
Trying 192.168.0.104... Connected to 192.168.0.104 (192.168.0.104). Escape character is '^]'. /opt/shell/mysqlchk_status_3306.sh: line 24: /tmp/processlist3306.txt: Permission denied /opt/shell/mysqlchk_status_3306.sh: line 25: /tmp/rep3306.txt: Permission denied HTTP/1.1 200 OK Connection closed by foreign host.
之前用root运行过所以报错,在被监控机删除临时文件
rm -f /tmp/processlist3306.txt /tmp/processlist3307.txt rm -f /tmp/rep3306.txt /tmp/rep3307.txt
没有输出则需检查mysqlchk_status_3306.sh脚本执行权限
启动后/var/log/messages 中会有很多日志
Oct 23 14:37:00 lova xinetd[11057]: START: mysqlchk_status3306 pid=11464 from=192.168.0.22 Oct 23 14:37:00 lova xinetd[11057]: EXIT: mysqlchk_status3306 status=0 pid=11464 duration=0(sec) Oct 23 14:37:05 lova xinetd[11057]: START: mysqlchk_status3306 pid=11494 from=192.168.0.22 Oct 23 14:37:05 lova xinetd[11057]: EXIT: mysqlchk_status3306 status=0 pid=11494 duration=0(sec)
在haproxy配置中将日志输出到黑洞
log_type = FILE /dev/null
查看监控
直接访问localhost是503
http://localhost/
503 Service Unavailable
No server is available to handle this request.
加上admin-status
http://localhost/admin-status
应用时需在slave mysql上的mysql添加通过haproxy的用户权限
haproxy的命令
/etc/init.d/haproxy
Usage: haproxy {start|stop|restart|reload|condrestart|status|check}
附
优化time_wait,防止端口耗尽
vi /etc/sysctl.conf
net.ipv4.ip_local_port_range = 1025 65000 net.ipv4.tcp_tw_reuse = 1 net.ipv4.tcp_tw_recycle = 1 net.ipv4.tcp_fin_timeout = 15 net.ipv4.tcp_max_tw_buckets = 35000
sysctl -p
使用nginx反向代理haprox后台
#省略
listen admin_status
mode http
bind 192.168.0.107:8000
option httplog
log global
stats enable
stats refresh 30s
stats hide-version
stats realm Haproxy\ Statistics
#stats uri /admin-status
stats uri /haproxy/
#stats auth admin:123456
#stats admin if TRUE
nginx.conf
#省略
location ~* ^/haproxy/
{
proxy_pass http://192.168.0.107:8000;
proxy_set_header Host $host;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
#proxy_set_header X-Forwarded-For $remote_addr;
proxy_redirect off;
}
#省略
参考:
http://linux.die.net/man/5/xinetd.conf
http://adslroot.blogspot.com/2013/12/haproxy-mysql.html
http://sssslide.com/www.slideshare.net/Severalnines/haproxy-mysql-slides