I have a program that forks a child, sleeps for a while and then kills the child and waits for it.
On CentOs (4) the wait sometimes fails with a "No child" error. I have not seen this on any other system. This program has been running for years on Solaris, older Redhat distributions, etc. I would say the failure rate is about 1 in 10.
I put a breakpoint on the error and forced the code to wait again. It works the second time.
Is this a known bug?
Mike
Michael Ubell wrote:
I have a program that forks a child, sleeps for a while and then kills the child and waits for it.
On CentOs (4) the wait sometimes fails with a "No child" error. I have not seen this on any other system. This program has been running for years on Solaris, older Redhat distributions, etc. I would say the failure rate is about 1 in 10.
I put a breakpoint on the error and forced the code to wait again. It works the second time.
Is this a known bug?
not known as far as I can tell. Is there a test script that can reproduce this issue ? is this on a specific Arch ?
can you file at http://bugs.centos.org/ - we can handle it there perhaps ?
- K
not known as far as I can tell. Is there a test script that can reproduce this issue ? is this on a specific Arch ?
This is intel x86_64. This is part of a larger program, I have not tried to minimize the the test case. I did notice that the child process has definitely gone zombie at the time it hits the breakpoint after the wait() fails.
I will submit a bug.
-
Karanbir Singh -
Michael Ubell