创建可fork子进程的C守护进程时出现僵尸进程问题求助

Hey there! Let's break down why your child processes are turning into zombies after setting up the daemon, and how to fix it.

What's Causing the Zombies? #

Looking at your ps -ajx output and code, the main culprit is likely incorrect error handling in your wait loop combined with potential unhandled failures in execv. Here's the breakdown:

• When wait() gets interrupted by a signal (like EINTR), your current code treats it as a child exit, decrements num_of_children, and eventually exits the loop. This leaves un-reaped child processes that become zombies.
• If execv fails (e.g., wrong path, missing execute permissions), the child process exits immediately—but if your wait loop isn't working right, it won't get reaped.

Step-by-Step Fixes #

1. Fix the wait Loop Error Handling #

Update your wait loop to properly handle interrupts and errors, so it only decrements the child count when a valid child exit is detected:

int status;
pid_t pid;
while(num_of_children > 0) {
    pid = wait(&status);
    if (pid == -1) {
        if (errno == EINTR) {
            // Wait was interrupted by a signal—keep waiting
            continue;
        } else if (errno == ECHILD) {
            // No more children to wait for, exit loop
            syslog(LOG_INFO, "No remaining child processes to monitor");
            break;
        } else {
            // Log other wait errors and exit
            syslog(LOG_ERR, "Wait failed: %s", strerror(errno));
            break;
        }
    }
    // Log valid child exit
    syslog(LOG_INFO, "Child %d exited with status 0x%x", pid, status);
    num_of_children--;
}

This ensures you don't incorrectly count interrupts as child exits, and properly handle all edge cases for wait().

2. Add Error Handling for execv #

If execv fails (returns instead of replacing the process), your child process will exit immediately—but you should log this failure to debug issues like missing executables:

if(i == 0) {
    syslog(LOG_INFO,"Forked message_server...");
    char * argv_list[] = {"message_queue_server",NULL};
    execv("./assignment_queue_server",argv_list);
    // If we reach here, execv failed
    syslog(LOG_ERR, "Failed to start message_queue_server: %s", strerror(errno));
    exit(EXIT_FAILURE);
}
if(i == 1) {
    syslog(LOG_INFO,"Forked time_check...");
    char * argv_list[] = {"time_check_process",NULL};
    execv("./time_check_process",argv_list);
    // If we reach here, execv failed
    syslog(LOG_ERR, "Failed to start time_check_process: %s", strerror(errno));
    exit(EXIT_FAILURE);
}

Check your system logs (usually /var/log/syslog or /var/log/messages) for these error messages if you still see issues.

3. Improve Daemonization (Optional but Recommended) #

Your current daemon setup misses a critical step: a second fork. This ensures your daemon isn't a session leader, preventing it from accidentally opening a terminal. Update your daemonization code:

int main() {
    int pid = fork();
    if(pid > 0) {
        printf("Parent process exiting..\n");
        exit(EXIT_SUCCESS); // No need to sleep—exit immediately
    } else if (pid == 0) {
        printf("First child process..\n");
        if(setsid() < 0) {
            exit(EXIT_FAILURE);
        }
        // Second fork to avoid being session leader
        pid = fork();
        if(pid > 0) {
            exit(EXIT_SUCCESS);
        } else if(pid == 0) {
            umask(0);
            if(chdir("/") < 0) {
                exit(EXIT_FAILURE);
            }
            // Close all file descriptors
            int x;
            for(x = sysconf(_SC_OPEN_MAX); x>=0;x--) {
                close(x);
            }
            openlog("assignment_daemon", LOG_PID|LOG_CONS,LOG_USER);
            syslog(LOG_INFO,"assignment_daemon started...");
            // Rest of your code to fork child processes goes here
        }
    }
    return 0;
}

This follows the standard Unix daemonization best practices.

Additional Troubleshooting Tips #

• Verify that ./assignment_queue_server and ./time_check_process exist and have execute permissions (run ls -l on them).
• Check your system logs to confirm if the child processes are starting successfully or failing to execute.
• Remove the sleep(10) in your initial parent process—it's unnecessary and can delay daemon startup.

内容的提问来源于stack exchange，提问作者Pony1487