Home >Operation and Maintenance >Linux Operation and Maintenance >How to use haproxy load balancing mysql
The content of this article is about how to use haproxy load balancing mysql. It has certain reference value. Friends in need can refer to it. I hope it will be helpful to you.
The system is a one-master, multiple-slave model. The separation of reading and writing is implemented at the application level. Insert, delete, update, and alter go to the master library, and select goes to the slave library. Then I did load balancing for the slave library and used keepalived to combine haproxy.
Without further ado, let’s get started.
System environment: centos7
Having said that, I don’t know when the official community website of haproxy was blocked... Alas, you can download the latest stable version of haproxy through this address.
# wget https://www.haproxy.org/download/1.8/src/haproxy-1.8.9.tar.gz
It is recommended that friends read the readme file before installation
# cat /usr/src/haproxy-1.8.9/README
Install haproxy, the prefix keyword PREFIX here specifies the installation directory, use are capital letters. Generally, for regular source code installation, add the option "--prefix" after the configure command to distinguish it
# tar -zxf haproxy-1.8.9.tar.gz -C /usr/src/ # cd /usr/src/haproxy-1.8.9/ # make TARGET=linux2628 USE_PCRE=1 USE_OPENSSL=1 USE_ZLIB=1 PREFIX=/usr/local/haproxy # make install PREFIX=/usr/local/haproxy
Create haproxy program user and configuration file directory
# useradd -M -s /sbin/nologin haproxy # mkdir /etc/haproxy
The haproxy configuration file does not provide a sample in this version of the package. It needs to be created manually.
# touch /etc/haproxy/haproxy.cfg # more /etc/haproxy/haproxy.cfg #global log /dev/log local0 info log /dev/log local1 notice pidfile /var/run/haproxy.pid maxconn 400000 user haproxy group haproxy nbproc 1 daemon #defaults mode http log global option http-server-close option redispatch retries 0 timeout http-request 10s timeout queue 1m timeout connect 10s timeout client 1m timeout server 1m timeout http-keep-alive 300s timeout check 10s maxconn 100000 listen mysql_select bind *:3306 mode tcp balance roundrobin #balance source #option mysql-check user haproxy timeout server 15s timeout connect 15s server mysql162 192.168.5.162:3306 check port 3306 inter 5000 fall 5 server mysql163 192.168.5.163:3306 check port 3306 inter 5000 fall 5 server mysql164 192.168.5.164:3306 check port 3306 inter 5000 fall 5 listen stats mode http bind 0.0.0.0:7979 stats enable stats hide-version stats uri /haproxy?hello stats realm Haproxy\ Statistics stats auth qkc:pwd@123 stats admin if TRUE
The default log of haproxy is It is output to the system's syslog, which is not very convenient to view. In order to better manage haproxy
logs, we generally define them independently in the production line. The definition method is as follows:
(1) Modify the log configuration options in haproxy.cfg, add the following configuration to global, and record the info and notice logs to different files.
log /dev/log local0 info log /dev/log local1 notice
(2) Modify rsyslog configuration
In order to facilitate management, haproxy-related configurations are independently defined in /etc/rsyslog.d/haproxy.conf. This part of the configuration records the info and notice logs to the /var/log/haproxy181/{haproxy181-info.log,haproxy181-notice.log} files respectively, where "& ~" means that after the log is written to the log file, rsyslog Stop processing this message. The syntax configured here is written in rainerscript script language.
# vim /etc/rsyslog.d/haproxy.conf if ($programname == 'haproxy' and $syslogseverity-text == 'info') then -/var/log/haproxy181/haproxy181-info.log& ~if ($programname == 'haproxy' and $syslogseverity-text == 'notice') then -/var/log/haproxy181/haproxy181-notice.log& ~
Check whether the syntax of haproxy.cfg configuration file is correct
# /usr/local/haproxy/sbin/haproxy -f /etc/haproxy/haproxy.cfg -c
If there is no problem, remove the -c option and start haproxy
# /usr/local/haproxy/sbin/haproxy -f /etc/haproxy/haproxy.cfg
[root@server181 haproxy]# ss -antulp | grep haproxy tcp LISTEN 0 128 *:3306 *:* users:(("haproxy",pid=3315,fd=4)) tcp LISTEN 0 128 *:7979 *:* users:(("haproxy",pid=3315,fd=6))
Configure haproxy to start automatically at boot
# chmod +x /etc/rc.d/rc.local # echo '/usr/local/haproxy/sbin/haproxy -f /etc/haproxy/haproxy.cfg' >> /etc/rc.d/rc.local
Connect haproxy for testing
# for i in $(seq 1 10); do mysql -utest -p123456 -h192.168.5.181 -e 'select @@server_id;'; done | egrep '[0-9]' 3306164 3306162 3306163 3306164 3306162 3306163 3306164 3306162 3306163 3306164
Then use another server to do haroxy and integrate keepalived. The remote system uses the mysql client to connect to the load balancing VIP to log in, and then perform health check, failover and other project tests one by one.
Summary:
haproxy has a very low CPU load on medium-sized loads, even in very high load scenarios, 5% user space occupancy and 95% system Space usage is also a very common phenomenon, which means that haproxy process consumption is more than 20 times lower than system space consumption. Therefore, performance tuning of the OS is very important. Even if the occupancy rate of user space is doubled, its CPU occupancy rate is only 10%, which also explains why layer 7 processing has limited impact on performance. As a result, haproxy's layer 7 performance can easily exceed hardware load balancing devices on high-end systems.
The feature of haproxy that supports tcp proxy allows it to also perform load balancing for mysql reading library, and it can perform quite well.
The above is the detailed content of How to use haproxy load balancing mysql. For more information, please follow other related articles on the PHP Chinese website!