| View previous topic :: View next topic |
| Author |
Message |
sulek
n00b
Joined: 04 Jun 2005
Posts: 55
|
 Posted: Tue Mar 21, 2006 8:55 pm Post subject: Can't kill (-9) process |
 |
|
Hello gyus!
I can't access my streamer (resource is busy) nor can I kill process that blocks it:
| Code: |
[root@server ~]# mt -f /dev/nst0 tell
/dev/nst0: Device or resource busy
[root@server ~]# ps aux|grep cpio
root 7170 1.4 0.0 4164 432 ? D Mar20 22:36 cpio -itv -I /dev/nst0
root 8581 0.0 0.0 5084 652 pts/0 S+ 21:49 0:00 grep cpio
[root@server ~]# kill -9 7170
[root@server ~]# kill -9 7170
[root@server ~]# ps aux|grep cpio
root 7170 1.4 0.0 4164 432 ? D Mar20 22:36 cpio -itv -I /dev/nst0
root 8583 0.0 0.0 4468 652 pts/0 S+ 21:49 0:00 grep cpio |
What can I do (please don't advise me to reboot server;)
Any help would be appreciated! |
|
| Back to top |
|
 |
Zwartoog
Tux's lil' helper
Joined: 13 Mar 2006
Posts: 87
Location: Netherlands
|
| Posted: Tue Mar 21, 2006 9:02 pm Post subject: |
|
|
The 'D' means the process is dead, i.e. not responding. The process is probably waiting for a device to respond or something.
There are 2 options: reset or wait for about a month or so (it is said that the process eventually gives up).
Dead processes are bad  |
|
| Back to top |
|
|
1clue
Advocate
Joined: 05 Feb 2006
Posts: 2569
|
| Posted: Tue Mar 21, 2006 9:07 pm Post subject: |
|
|
I had a similar issue on AIX once.
What was happening is that something being called by /etc/inittab was taking a long time to execute. As a result, the init process could not get around to do housekeeping.
I had what were called zombie processes hanging around, which were processes that had terminated but had not been cleaned up by the OS.
This was self-inflicted. I knew everything about /etc/inittab since I had read the man page (!) and started hacking my own scripts in. This issue took me a while to figure out, but I suddenly realized that having read a man page does not make one an expert on the underpinnings of unix. |
|
| Back to top |
|
|
bollucks
l33t
Joined: 27 Oct 2004
Posts: 606
|
| Posted: Thu Mar 23, 2006 1:14 am Post subject: |
|
|
| D means sleeping in uninterruptible sleep. This means it is waiting for some kernel process to finish on its behalf and return... and it hasn't returned. There is no way to kill such a task. It means there has been a bug in some kernel process, NFS on removed mounts being the most common scenario. Z means zombie and it means the process wasn't responsible for cleaning up after itself but the parent, and the parent has not yet cleaned up. |
|
| Back to top |
|
|
1clue
Advocate
Joined: 05 Feb 2006
Posts: 2569
|
| Posted: Thu Mar 23, 2006 1:44 am Post subject: |
|
|
A bug in the kernel process is not really necessary for that. What it means is the kernel is tied up doing something lengthy. That could be a bug, or it could be some clever script you run from /etc/inittab. /etc/inittab is run as the kernel process, as are the files referenced by it. All I had to do is "&" the script I was tinkering with, in addition to piping output to a file, and my zombies all went away.
On AIX, anyway, a process that was killed turned into a zombie, and the init process cleaned it up after that. Generally it happens so fast you never see a zombie process, but in the event of a hanging or lengthy task in /etc/inittab you DO see them.
I can't say where the lengthy process is in this case if it isn't a bug, but I'd start looking at any custom scripts that were created recently. |
|
| Back to top |
|
|
|
Kernel & Hardware
