Existing implementations vary on the result of a kill() with pid indicating an inactive process (a terminated process that has not been waited for by its parent). Some indicate success on such a call (subject to permission checking), while others give an error of [ESRCH]. Since the definition of process lifetime in this volume of POSIX.1-2008 covers inactive processes, the [ESRCH] error as described is inappropriate in this case. In particular, this means that an application cannot have a parent process check for termination of a particular child with kill(). (Usually this is done with the null signal; this can be done reliably with waitpid().
Regards,
Alister
Historic implementations in 1988 (when the first System Interfaces volume of the POSIX standards was approved) behaved both ways. (Notably UNIX System V succeeded, and 4.3BSD returned an ESRCH error.) The standard required the System V behavior, and that was further reinforced when the 2008 edition of the standard clarified that the lifetime of a process does not end until it is reaped.
But, of course, some implementations of UNIX-like systems do not conform to the standards.
Zombie processes which are not cleaned up by unix within a few hours can really only be cleared by a reboot.
The odd few zomboes will be caused by untidy disconnects from clients. If you get large numbers, suspect a hardware or software fault. After replacing a failed hot-pull tape drive it could still be necessary to reboot the computer to clear zombie processes left with incomplete I/O from the original fault. I have seen faulty client/server software which accumulated zombies at an alarming rate.
Your post does not describe zombies (processes which have already exited), but living processes which are permanently stuck in an uninterruptible sleep (as you mentioned, typically due to a driver bug and/or hardware fault).
I had a problem deleting a zombie process. It refused to be killed.
I even tried kill -9 process# but it refused.
Any other way of killing it? (7 Replies)
i'm writing small http proxy server (accept client -> connect to remote proxy server -> recv client's request -> send to remote proxy server -> get responce from remote proxy server -> send answer to client -> close connection to client and to remote proxy server) and having problems with fork().... (2 Replies)
Hi All
I need help, how can i kill zombies instead of rebooting the system.
Regards
System: sna Tue Apr 5 17:50:23 2005
Load averages: 0.05, 0.15, 0.22
168 processes: 157 sleeping, 5 running, 6 zombies
Cpu states:
CPU LOAD USER NICE... (5 Replies)
Okay, I'm working within ansi C and Sun Solaris 7. I have a problem with zombies. I'm currently using the kill command to return the status of a process. How do I check for Zombie PIDs or the right function to return its PID from within a C program? (1 Reply)