Stubborn Processes

[Steven Erat]

> Or zombie processes.

Thanks for the reply.  <defunct> does not appear in the process list, so I
don't believe they are zombie.  

I'd still like to understand the conditions required and the mechanism to
make a process resistant to being killed by root. 

I'm still waiting on ps -elf | grep <processname> and strace from the
client.  I was running strace myself against my own copy of the appserver,
and I actually never saw it output any data even when handling requests. 
Whenever I use truss on solaris it outputs with much verbosity, so I
thought perhaps I was stracing incorrectly. Here's what I got after
attaching strace to the parent process of the server, then making some
requests, and then CTRL-C:

[root at merlot /root]# strace -p 11988
pause(
 <unfinished ...>
[root at merlot /root]#

Is this right?  Should I use a different combinary of switches?

Thanks,
Steven Erat


--- ron.peterson at yellowbank.com wrote:
> On Fri, Jul 18, 2003 at 10:00:23AM -0400, markw at mohawksoft.com wrote:
> > 
> > >> My question is how can it be that root cannot kill a process?  
> > 
> > If root can not kill a process, that means the process is in a
> condition in
> > which it can not be killed. (Nice reasoning huh?)
> > 
> > Actually, the only time I have seen this is when a process is hung in
> a
> > device driver.
> 
> Or zombie processes.
> 

__________________________________
Do you Yahoo!?
SBC Yahoo! DSL - Now only $29.95 per month!
http://sbc.yahoo.com