上一篇文章搭建了plg日志监控，本次准备看下目前集群的性能。
对应机器如下：虚机4，前一后三，4台配置均为4核4g
物理机为20核32g台式机。
先进行原始nginx配置的压测 nginx配置如下：
user nginx;
worker_processes 1;
error_log /var/log/nginx/error.log warn;
pid /var/run/nginx.pid;
events {
worker_connections 1024;
}
http {
include /etc/nginx/mime.types;
default_type application/octet-stream;
#配置nginx json格式日志
log_format main '{"@timestamp": "$time_local", '
'"remote_addr": "$remote_addr", '
'"referer": "$http_referer", '
'"request": "$request", '
'"status": $status, '
'"bytes": $body_bytes_sent, '
'"agent": "$http_user_agent", '
'"x_forwarded": "$http_x_forwarded_for", '
'"up_addr": "$upstream_addr",'
'"up_host": "$upstream_http_host",'
'"up_resp_time": "$upstream_response_time",'
'"request_time": "$request_time"'
' }';
access_log /var/log/nginx/access.log main;
sendfile on;
#tcp_nopush on;
keepalive_timeout 65;
#gzip on;
upstream test.miaohr.com {
server 192.168.175.128:8070;
server 192.168.175.129:8070;
server 192.168.175.130:8070;
}
server {
listen 80;
server_name test.miaohr.com;
charset utf-8;
location / {
root html;
index index.html index.htm;
proxy_pass http://test.miaohr.com;
proxy_set_header X-Real-IP $remote_addr;
client_max_body_size 100m;
}
}
}
使用jmeter5.0进行压测，压测命令如下
jmeter -n -t D:/testYc.jmx -l D:/result.txt -e -o D:/webreport
其中
D:/testYc.jmx ------> 测试计划文件的路径
D:/result.txt ------> 将要生成的测试结果文件的存放路径
D:/webreport -------> 将要生成的web报告的保存路径
使用物理机cmd命令进行压测。线程为200，持续100s，报告如下
此时rt为180ms+,tps在980左右
修改nginx配置为长连接，长连接配置如下
user nginx;
worker_processes 1;
error_log /var/log/nginx/error.log warn;
pid /var/run/nginx.pid;
events {
worker_connections 1024;
}
http {
include /etc/nginx/mime.types;
default_type application/octet-stream;
#配置nginx json格式日志
log_format main '{"@timestamp": "$time_local", '
'"remote_addr": "$remote_addr", '
'"referer": "$http_referer", '
'"request": "$request", '
'"status": $status, '
'"bytes": $body_bytes_sent, '
'"agent": "$http_user_agent", '
'"x_forwarded": "$http_x_forwarded_for", '
'"up_addr": "$upstream_addr",'
'"up_host": "$upstream_http_host",'
'"up_resp_time": "$upstream_response_time",'
'"request_time": "$request_time"'
' }';
access_log /var/log/nginx/access.log main;
sendfile on;
#tcp_nopush on;
keepalive_timeout 65;
#gzip on;
upstream test.miaohr.com {
server 192.168.175.128:8070;
server 192.168.175.129:8070;
server 192.168.175.130:8070;
keepalive 500;#最大空闲连接数
}
server {
listen 80;
server_name test.miaohr.com;
charset utf-8;
location / {
root html;
index index.html index.htm;
proxy_pass http://test.miaohr.com;
proxy_set_header X-Real-IP $remote_addr;
client_max_body_size 100m;
}
}
location / {
proxy_pass http://test-api;
proxy_set_header Host $host;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header REMOTE-HOST $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_http_version 1.1; # 设置http版本为1.1
proxy_set_header Connection ""; # 设置Connection为长连接（默认为no）
}
}
进行第二次压测，报告如下
本次压测，rt降低到76ms，tps上升到2400+。对比短连接，集群的性能得到了很大的提升。但错误率指标有所上升。根据jmeter执行日志查看， 报错集中为以下内容
1650551535574,485,HTTP请求,Non HTTP response code: java.net.BindException,Non HTTP response message: Address already in use: connect,线程组 1-135,text,false,,2437,0,200,200,http://192.168.175.131/getSxbm,0,0,485
根据查阅后发现，这是由于压测机为windows系统，tcp端口资源来不及释放导致。与集群性能无关
长连接与短连接的区别，主要在于长连接模式下，不需要频繁的建立起tcp连接，同时也不需要频繁关闭，节省了这部分资源的消耗。这种情况下比较适合频繁请求的场景下。
