Hello I am having serious trouble with the fork command, i basically want to create 9 or 10 child processes and store their pid numbers in array while the children stay resident until i kill() them later , i cannot seem to control how many are made as they all seem to create their own children.
I have managed to stop crashing my system, can somebody help please, none of my books are any use.
This is a very common problem observed with fork.
You can have some flag variable which will prevent the child process to fork again when the parent forks.
void main()
{
int child1, child2, val, if_child;
char err[1000];
if_child = 1;
memset(err,'\0',strlen(err));
printf("\nParent process ID is %d \n",getpid());
child1 = fork();
if (child1 == -1)
{
strcpy(err, strerror(errno));
}
else
{
if(child1 > 0) /* If fork command is successful child PID will be greater than zero. */
{
printf("\nThe child1 process ID is %d \n", child1);
if_child = 0;
/* Set the flag here to prevent child process from forking */
}
}
memset(err,'\0',strlen(err));
if (if_child == 0)
{
child2 = fork();
if (child2 == -1)
{
strcpy(err, strerror(errno));
}
printf("\nThe child2 process ID is %d \n", child2);
Thanks a lot, looks promising, i look forward to the day when i can answer someone else's *NIX programmig question. Looks like it's gonna take longer than with VB.
I had the same problem with one of my projects (you can see it if the thread called 'making a process tree') and getting more child processes than i expected was really a nasty problem.
Still haven't figured out how it works exactly but i hope aniruddha's advice will help me too.
Yep. That's one way to do it or you could just send a stop signal so the child processes won't waste your memory for nothing. Just replace for(;;)
with kill(getpid(),SIGSTOP).
That's all fine and dandy but what if you want the child process to actually do something? Cause if you stop it or send it in an infinite loop then you can't quite work within the child can you?
That's actually the part i'm interested in: working within the child but prevent it from forking once the parent forks again.
You guys are making this much harder than it needs to be. Let's start by forking one process and storing its pid:
Code:
#include <stdio.h>
#include <unistd.h>
void main()
{
int pid, parentpid, childpid;
parentpid=getpid();
printf("I am the parent process and my pid is %d\n", getpid());
if (pid=fork()) {
childpid=pid;
} else {
printf("I am a child process and my pid is %d\n", getpid());
exit(0);
}
printf("I am still the parent process and my pid is %d\n", getpid());
exit(0);
}
Here the child process just displays its pid then exits. You will probably want to do more with your child processes, but after your children take care of business they must exit so they participate in any further forking. Also in my example the parent process exits fairly quickly. This means that init will inherit the child process and will reap it when it dies. If I wanted to keep the parent around, I would need to insure that it issues wait() calls for each child who dies. If I didn't do this, the children would become zombies. I usually just let the parent die.
Once we have some code that does what we want, if we want to do it n times, we use a loop:
Code:
#include <stdio.h>
#include <unistd.h>
void main()
{
int n, pid, parentpid, childpids[10];
parentpid=getpid();
printf("I am the parent process and my pid is %d\n", getpid());
for(n=0; n<5; n++) {
if (pid=fork()) {
childpids[n]=pid;
} else {
printf("I am a child process and my pid is %d\n",
getpid());
exit(0);
}
}
printf("I am still the parent process and my pid is %d\n", getpid());
exit(0);
}
As requested by the OP, the children's pids are recorded in an array. But I still just let the parent die.
I am writing a bash script to automate the installation of web environment on a base install of Fedora. And I'm at the limit of my last nerve and my bash skills. My brain is screaming at me: "Give up and use perl", but I am trying to stick to bash since the script will modify the perl environment... (6 Replies)
I have a file that is 20 - 80+ MB in size that is a certain type of log file.
It logs one of our processes and this process is multi-threaded. Therefore the log file is kind of a mess. Here's an example:
The logfile looks like: "DATE TIME - THREAD ID - Details", and a new file is created... (4 Replies)
I don't want to speak about the goods or bads of both kinds of Operating systems, I only want to share a little experience with you to comment it.
I live in Spain and I have home some old unix systems, some of them that I want to sell or change for other things, like a pair of Sun Blade 2000... (0 Replies)
I've been able to generate output based on the code scarfake provided me (thanks again man).
A little background so everyone more or less knows whats going on:
I needed code that would propagate a database with 100,000 entries, for capacity testing purposes, something like a stress test.
... (5 Replies)
Hi everyone,
I have a requirement that requires me to fill an sqlite database with 100,000 entries (no duplicates).
I will start out by giving the command that will insert the values necessary to populate the database:
# sqlite /var/local/database/dblist "insert into list... (2 Replies)
I've just installed redhat 6.2 on one of my systems and am trying to install the gcc c compiler after downloading an rpm from the redhat site. The damn thing gives me:
only major numbers <= 3 are supported by this version of RPM
what do I do, it does the same with the latest rpm of php
... (7 Replies)
fork()ing hell!!