1. OBPROXY 基于TCP keep-alive的超時(shí)檢測(cè)和保活機(jī)制

• 在分析某數(shù)據(jù)包時(shí)，我們發(fā)現(xiàn)，針對(duì)空閑的TCP連接，obproxy 服務(wù)端每隔5分鐘都會(huì)發(fā)送一些特殊的 tcp數(shù)據(jù)包，這些數(shù)據(jù)包在wireshark中顯示為 [TCP KEEP-ALIVE]，其示例如下：：
• 這些數(shù)據(jù)包其實(shí)就是大名鼎鼎的tcp keep-alive 心跳包。而 obproxy 之所以會(huì)發(fā)送這些心跳包，其實(shí)是因?yàn)閛bproxy 開(kāi)啟了其SO_KEEPALIVE 選項(xiàng)(keep-alive packets are sent only when the SO_KEEPALIVE socket option is enabled)，并使用了 LINUX 操作系統(tǒng)的  socket 套接字級(jí)別的基于 tcp keep-alive的超時(shí)檢測(cè)和?；顧C(jī)制，該機(jī)制的詳情見(jiàn)后文。
• obproxy相關(guān)參數(shù)和配置方式如下：

alter proxyconfig set server_tcp_user_timeout = 0;
alter proxyconfig set sock_option_flag_out = 3;
alter proxyconfig set server_tcp_keepidle = 5;
alter proxyconfig set server_tcp_keepintvl = 5;
alter proxyconfig set server_tcp_keepcnt = 2;
alter proxyconfig set client_sock_option_flag_out = 3;
alter proxyconfig set client_tcp_keepidle = 5;
alter proxyconfig set client_tcp_keepintvl = 5;
alter proxyconfig set client_tcp_keepcnt = 2;
alter proxyconfig set client_tcp_user_timeout = 0;
alter proxyconfig set skip_proxyro_check=true;

2. LINUX基于TCP keep-alive的超時(shí)檢測(cè)和保活機(jī)制

Linux操作系統(tǒng)中，基于TCP keep-alive的超時(shí)檢測(cè)和?；顧C(jī)制，分為兩個(gè)層面，一個(gè)是操作系統(tǒng)級(jí)別的，一個(gè)是 socket 套接字級(jí)別的。

2.1. LINUX中操作系統(tǒng)級(jí)別的基于TCP keep-alive的超時(shí)檢測(cè)和?；顧C(jī)制

操作系統(tǒng)級(jí)別的，基于TCP keep-alive的超時(shí)檢測(cè)和?；顧C(jī)制，主要跟以下幾個(gè)內(nèi)核參數(shù)相關(guān)，可以在操作系統(tǒng)層面，通過(guò) sysctl 命令查看和更改這些內(nèi)核參數(shù)：

• /proc/sys/net/ipv4/tcp_keepalive_intvl: 默認(rèn) 75秒，The number of seconds between TCP keep-alive probes；
• /proc/sys/net/ipv4/tcp_keepalive_probes: 默認(rèn) 9 次，The maximum number of TCP keep-alive probes to send before giving up and killing the connection if no response is obtained from the other end；
• /proc/sys/net/ipv4/tcp_keepalive_time: 默認(rèn) 7200 秒即2小時(shí)，The number of seconds a connection needs to be idle before TCP begins sending out keep-alive probes. Keep-alives are sent only when the SO_KEEPALIVE socket option is enabled. An idle connection is terminated after approximately an additional 11 minutes (9 probes an interval of 75 seconds apart) when keep-alive is enabled；
• sysctl net.ipv4.tcp_keepalive_time
• sysctl net.ipv4.tcp_keepalive_intvl
• sysctl net.ipv4.tcp_keepalive_probes

2.2. LINUX中socket套接字級(jí)別的基于TCP keep-alive的超時(shí)檢測(cè)和?；顧C(jī)制

socket 套接字級(jí)別的，基于TCP keep-alive的超時(shí)檢測(cè)和?；顧C(jī)制，則需要相關(guān)應(yīng)用在其代碼中，指定如下這些 socket 套接字選項(xiàng)，事實(shí)上 obproxy 就是利用了該機(jī)制：

• TCP_KEEPIDLE：the amount of time until the first keepalive packet is sent；
• TCP_KEEPCNT：the number of probes to send；
• TCP_KEEPINTVL：the interval between keepalive packets；

3. JAVA中如何指定 socket端口級(jí)別的基于TCP keep-alive的超時(shí)檢測(cè)和?；顧C(jī)制

JDK11及之后的版本，也支持socket端口級(jí)別的，基于TCP keep-alive的超時(shí)檢測(cè)和保活機(jī)制配置，事實(shí)上大部分 JDK8 版本，也在代碼層面通過(guò) backport 支持了該機(jī)制，相關(guān)源碼如下：

• java.net.StandardSocketOptions
• java.net.StandardSocketOptions#SO_KEEPALIVE
• jdk.net.ExtendedSocketOptions
• When the SO_KEEPALIVE option is enabled, TCP probes a connection that has been idle for some amount of time. The default value for this idle period is 2 hours which is too long for most applications. The TCP_KEEPIDLE, TCP_KEEPCOUNT, TCP_KEEPINTERVAL option can be used to affect this value for a given socket.
• The default idle time for SO_KEEPALIVE is 2 hours, too long for most applications. Some operation systems have support to configure the idle time on a per connection basis (Linux has TCP_KEEPIDLE, Windows has SIO_KEEPALIVE_VALS). We should consider exposing an extended socket option to configure this.
• TCP_KEEPIDLE, TCP_KEEPCOUNT, and TCP_KEEPINTERVAL are non-standard socket options supported on several platforms to provide fine control over the TCP/IP keep alive mechanism. It should be possible to set these socket options via the setOption method defined by java.net.Socket and java.nio.channels.SocketChannel.
• Add a JDK-specific socket option that supports setting TCP_KEEPIDLE, TCP_KEEPCOUNT, TCP_KEEPINTERVAL, on platforms that support it. The option can be set/get through the existing set/getOption methods on Socket and NetworkChannel.

圖片

圖片

注意：如果執(zhí)行 JAVA 程序時(shí)，遇到如下錯(cuò)誤，Exception in thread "main" java.lang.NoSuchFieldError: TCP_KEEPIDLE，通常是因?yàn)槭褂玫?JDK 版本不支持 TCP_KEEPIDLE等jdk.net.ExtendedSocketOptions 擴(kuò)展選項(xiàng)， 這些擴(kuò)展選項(xiàng)是 Java 9 才正式引入的，只有部分版本的Java 8 支持該選項(xiàng)。

4. 參考鏈接

• https://bugs.openjdk.org/browse/JDK-8203194
• https://bugs.openjdk.org/browse/JDK-8194298
• https://bugs.openjdk.org/browse/JDK-8236187
• https://open.oceanbase.com/blog/10900349
• https://www.oceanbase.com/knowledge-base/oceanbase-database-1000000000763298
• https://segmentfault.com/a/1190000042448986/en