记一次Springboot集成druid系统间歇性报communications link failure

问题描述

springBoot项目集成druid的时候，系统间歇性的报CommunicationsException:communications link failure的异常如下。由于是间歇性的，即并不是每次都报异常，而是过了一段时间之后就会出现，所以针对性的测试修复难度很大。

2021-06-11T10:52:09.365Z  APP  2021-06-11 10:52:09.358 ERROR 1 --- [nio-8082-exec-2] c.a.d.p.DruidDataSource                  : discard connection
2021-06-11T10:52:09.365Z  APP  
2021-06-11T10:52:09.365Z  APP  com.mysql.cj.jdbc.exceptions.CommunicationsException: Communications link failure
2021-06-11T10:52:09.365Z  APP  
2021-06-11T10:52:09.365Z  APP  The last packet successfully received from the server was 300,811 milliseconds ago.  The last packet sent successfully to the server was 300,812 milliseconds ago.
2021-06-11T10:52:09.365Z  APP   at com.mysql.cj.jdbc.exceptions.SQLError.createCommunicationsException(SQLError.java:174) ~[mysql-connector-java-8.0.15.jar!/:8.0.15]
2021-06-11T10:52:09.365Z  APP   at com.mysql.cj.jdbc.exceptions.SQLExceptionsMapping.translateException(SQLExceptionsMapping.java:64) ~[mysql-connector-java-8.0.15.jar!/:8.0.15]
2021-06-11T10:52:09.365Z  APP   at com.mysql.cj.jdbc.ConnectionImpl.setReadOnlyInternal(ConnectionImpl.java:2179) ~[mysql-connector-java-8.0.15.jar!/:8.0.15]
2021-06-11T10:52:09.365Z  APP   at com.mysql.cj.jdbc.ConnectionImpl.setReadOnly(ConnectionImpl.java:2163) ~[mysql-connector-java-8.0.15.jar!/:8.0.15]
2021-06-11T10:52:09.365Z  APP   at com.mysql.cj.jdbc.ha.MultiHostMySQLConnection.setReadOnly(MultiHostMySQLConnection.java:520) ~[mysql-connector-java-8.0.15.jar!/:8.0.15]
2021-06-11T10:52:09.365Z  APP   at sun.reflect.GeneratedMethodAccessor198.invoke(Unknown Source) ~[?:?]
2021-06-11T10:52:09.365Z  APP   at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) ~[?:1.8.0_281]
2021-06-11T10:52:09.365Z  APP   at java.lang.reflect.Method.invoke(Method.java:498) ~[?:1.8.0_281]
2021-06-11T10:52:09.365Z  APP   at com.mysql.cj.jdbc.ha.FailoverConnectionProxy.invokeMore(FailoverConnectionProxy.java:530) ~[mysql-connector-java-8.0.15.jar!/:8.0.15]
2021-06-11T10:52:09.365Z  APP   at com.mysql.cj.jdbc.ha.MultiHostConnectionProxy.invoke(MultiHostConnectionProxy.java:479) ~[mysql-connector-java-8.0.15.jar!/:8.0.15]

经过研究一段时间之后，发现配置文件中的两个配置要注意。test-while-idle 和 time-between-eviction-runs-millis，具体配置的说明如下。

#druid连接池配置
spring:
  datasource:
    db:
      type: com.alibaba.druid.pool.DruidDataSource
      driver-class-name: com.mysql.jdbc.Driver    #高版本使用 com.mysql.cj.jdbc.Driver
      url: jdbc:mysql://localhsot:3306/db?autoReconnect=true&useSSL=false&failOverReadOnly=false&maxReconnects=10
      username: xxxx
      password: xxxx
    druid:
      max-active: 100                #指定连接池中最大的活跃连接数.
      initial-size: 10               #指定连接的初始值
      min-idle: 10                   #指定必须保持连接的最小值
      max-wait: 60000                #指定连接池等待连接返回的最大等待时间，设置1分钟；默认-1不限时间
      test-on-borrow: false          #获取连接时候验证，会影响性能，默认为false
      test-while-idle: true          #验证连接的有效性
      time-between-eviction-runs-millis: 300000     #空闲连接回收的时间间隔，与test-while-idle一起使用，设置5分钟
      min-evictable-idle-time-millis: 1800000       #连接池空闲连接的有效时间 ，设置30分钟
      validation-query: select 1
      remove-abandoned-timeout: 30   #隔30秒回收断开的连接
      remove-abandoned: true         #当连接超过了removeAbandonedTimout时间，删除泄露的连接，默认false
      log-abandoned: true            #当Statement或连接被泄露时打印程序的stack traces日志
      filter:
        slf4j:
          enabled: true                    #开启slf4j debug日志打印
          statement-log-enabled: false     #关闭statement相关debug日志打印
          result-set-log-enabled: false    #关闭result-set相关debug日志打印

如果想在开发环境复现这种问题的话，
首先设置mysql变量
set global interactive_timeout=30
set global wait_timeout=30;　（超时时间由28800改为30秒）
然后再把项目druid的配置文件test-while-idle=false
最后启动项目，第一次请求sql才会初始化连接池，等待30+秒，再次请求sql就会出现以上错误（CommunicationsException异常也不是只有一种情况，这里解决的是java.io.EOFException: Can not read response from server. Expected to read 4 bytes, read 0 bytes before connection was unexpectedly lost.