集群下的Failover Replication笔记

集群环境下，failover replication的理解：
HTTP proxy接收到某一http请求后，会根据配置的load balance算法选择一个节点做为主节点去执行，并把这个HTTP请求的session复制保存在backup节点上。
1. 当主节点执行失败，如有异常，则把这个http请求重定向到backup节点上去重新执行。
2. 当主节点执行没有异常，但执行时间超过某个配置的值，则cluster也会认为这个节点failed （timeout），把这个请求重定向到backup节点上重新执行。配置是在Apache目录下的httpd.conf中，default value is 300 second。

<VirtualHost 1.2.3.4:80>
        ServerAdmin [e-mail]
        DocumentRoot /abacus/farex
        ServerName 1.2.3.4
        ErrorLog "|/opt/apache2.2/bin/rotatelogs /abacus/farex/logs/farex_error_log.%Y%m%d 86400"
        CustomLog "|/opt/apache2.2/bin/rotatelogs /abacus/farex/logs/farex_access_log.%Y%m%d 86400" common
        <Location / >
                SetHandler weblogic-handler
                WebLogicCluster 1.2.3.4:1111,1.2.3.5:1111
                WLIOTimeoutSecs 86400
                WLSocketTimeoutSecs 30
                ConnectRetrySecs 300
                ConnectTimeoutSecs 300
                KeepAliveEnabled ON
                KeepAliveSecs 60
                MaxPostSize -1
                Debug ALL
                DebugConfigInfo OFF
                WLTempDir '/abacus/farex/logs'
                WLLogFile "|/opt/apache2.2/bin/rotatelogs /abacus/farex/logs/wlproxy.log.%Y%m%d 86400"
                ErrorDocument 404 "   xxxxx7 says: Fare Update Servers not available (404)"
        </Location>
</VirtualHost>