unix 设置保活机制

[thomaswang]

int keepAlive = 1;   // 开启 keepalive 属性. 缺省值: 0(关闭)  
int keepIdle = 60;   // 如果在 60 秒内没有任何数据交互,则进行探测. 缺省值:7200(s)  
int keepInterval = 5;   // 探测时发探测包的时间间隔为 5 秒. 缺省值:75(s)  
int keepCount = 2;   // 探测重试的次数. 全部超时则认定连接失效..缺省值:9(次)  
setsockopt(s, SOL_SOCKET, SO_KEEPALIVE, (void*)&keepAlive, sizeof(keepAlive));  
setsockopt(s, SOL_TCP, TCP_KEEPIDLE, (void*)&keepIdle, sizeof(keepIdle));  
setsockopt(s, SOL_TCP, TCP_KEEPINTVL, (void*)&keepInterval, sizeof(keepInterval));  
setsockopt(s, SOL_TCP, TCP_KEEPCNT, (void*)&keepCount, sizeof(keepCount));  

这个是 server 端还是 client 端设置，还是都设置？

[jimzhong]

LZ 看一下 http://www.tldp.org/HOWTO/html_single/TCP-Keepalive-HOWTO/吧

[Panic]

保活一般都是应用自己加的，基本没人用 tcp 层面的 KEEPALIVE

[thomaswang]

@Panic 能说说自己怎么加吗，服务端(对每个连接)、客户端都要设置计时，到时间就发个包给对放 ？

[Panic]

@thomaswang 不叫保活, 不都叫心跳吗, 反正我们自己都是用心跳的. 类似下面这种
https://segmentfault.com/a/1190000006832547