Socket programming practices (8)

How to Avoid zombie processes by using multiple processes on the server

1) Solve botnets by ignoring sigchld Signals

Add

signal(SIGCHLD, SIG_IGN);

2) Solve botnets through the wait/waitpid Method

　　signal(SIGCHLD,onSignalCatch);　　　　void onSignalCatch(int signalNumber)　　{　　    wait(NULL);　　}

Server code:

// Return a socket that have start listened.int mkatcpserver (int serverport, int backlog = somaxconn) {int sockfd = socket (af_inet, sock_stream, 0); If (sockfd =-1) {err_exit ("socket error");} // Add address reused int on = 1; if (setsockopt (sockfd, sol_socket, so_reuseaddr, & on, sizeof (on )) =-1) {err_exit ("setsockopt so_reuseaddr error");} // band a local address and port struct sockaddr_in serveraddr; serveraddr. sin_family = af_inet; serveraddr. sin_port = htons (SERVERPORT); serveraddr. sin_addr.s_addr = inaddr_any; // band an any IP address if (BIND (sockfd, (struct sockaddr *) & serveraddr, sizeof (serveraddr) =-1) {err_exit ("BIND error");} // start to listen. if (Listen (sockfd, backlog) =-1) {err_exit ("Listen error");} return sockfd;} void onsignalcatch (INT signalnumber) {waitpid (-1, null, wnohang);} int main () {// install sigchld signal processing function signal (sigchld, onsignalcatch); int serversockfd = mkatcpserver (8002); struct sockaddr_in peeraddr; socklen_t peerlen = sizeof (peeraddr); While (true) {// accept the link int peersockfd = accept (serversockfd, (struct sockaddr *) & peeraddr, & peerlen ); if (peersockfd =-1) {err_exit ("Accept error ");}..... // other code is similar to the previous one

Disable multiple clients at the same time

The problem is described as follows:

Client code implementation code

// Other Code such as // ...int main () {int serversocket [10]; int socketcount = 10; for (INT I = 0; I <socketcount; ++ I) {serversocket [I] = mkatcpclient (8002, "127.0.0.1");} Sleep (100); Return 0 ;}

When you press Ctrl + C during the running process, you can see that 10 sub-processes are started on the server and all clients are disconnected, the number of zombie processes generated is amazing (It also proves that the sigchld signal is unreliable )!

 

Solution:

// Some server code is as follows: // others such as the previous... void onsignalcatch (INT signalnumber) {int ret = 0; // pay attention to this !!!! While (ret = waitpid (-1, null, wnohang )! =-1);} int main () {signal (sigchld, onsignalcatch); int serversockfd = mkatcpserver (8002 );...

We can see that the server process has been monitored and no zombie process has been found!

Socket programming practices (8)