Summary: Killing a process with rapidly changing pid (without rebooting)

Thanks to the numerous replies.

====================================

Below are 5 which are good enough :

Use ptree to find its (stable) parent and then kill that, or
use pkill <script_name>.

====================================

I dont think that this solve you problem but:
kill -9 ` ps -elf | grep foo | grep -v "grep" | cut -d " " -f5`


====================================

If the name of the process is unique, you can use pkill:

pkill foobar

or

pkill -9 foobar

====================================

#ps -elf | grep nohup | grep -v grep
 0 O     root 16538  1579   0  99 20        ?    152          13:24:49 ?
0:00 nohup echo AA


In my output you see the parent process id is 1579. Now I am killing
this process:
#kill 1579

====================================

pgrep PARENT_PROCESS_NAME | xargs kill -9



Thanks
U

On Fri, Jul 18, 2008 at 2:55 PM, sunhux G <sunhux@gmail.com> wrote:

> Hi,
>
>
>
> I've started/run a "while ... loop" shell script by issuing
> "nohup script_name &".
>
> Rebooting the server is not an option.
>
> Problem is in a split of a second after issuing
>    "ps -ef | grep script_name"
> the pid would change again & I'm not able to kill using pid.
>
> I wrote this script so that I can rapidly check a SAN LUN
> path status & email out when the status of the SAN LUN
> changed momentarily.  We suspect there's transient/quick
> transitioning problem with certain SAN LUN & the SAN
> vendor told us to check it by issuing a specific command
> from the Solaris server.
>
> I did not insert a "sleep ..." line in the script in case during
> the short pause, we missed capturing the state of the SAN
> LUN.
>
>
> Perhaps someone has script that could quickly get the
> pid of the process, pipe it immediately to the kill command??
>
>
> Thanks
> U
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers