dtrace discuss - Apr 2006 - Control-c

why do I need to use Control-c to stop my trace, for example:

#dtrace -n ''profile:::tick-1sec/++x == 10/ {exit(0)}''

in my system with heavy IO load, I have to use Control-c

without heavy IO, the same trace cmd just return automatically after 10 sec. 

I just assume Control-c is to deliver a signal to dtrace, but why do I have to
in some cases (like I have a heavy IO load)?

Thanks
 
 
This message posted from opensolaris.org

I hope I can get some answers whatsoever to this questions. Is it a dumb or
stupid Q? it is 100% reproducible in my system T2000. I really appreciate for
any input.
 
 
This message posted from opensolaris.org

How long are you waiting before hitting ^C? If the dtrace(1M) can''t
check in
with the kernel every so often, it should abort on its own. Does truss(1)ing
it show anything interesting?

Adam

On Mon, Apr 10, 2006 at 03:04:36PM -0700, Qiang Liu
wrote:
> why do I need to use Control-c to stop my trace, for example:
> 
> #dtrace -n ''profile:::tick-1sec/++x == 10/ {exit(0)}''
> 
> in my system with heavy IO load, I have to use Control-c
> 
> without heavy IO, the same trace cmd just return automatically after 10
sec.
> 
> I just assume Control-c is to deliver a signal to dtrace, but why do I have
to in some cases (like I have a heavy IO load)?
> 
> Thanks
>  
>  
> This message posted from opensolaris.org
> _______________________________________________
> dtrace-discuss mailing list
> dtrace-discuss at opensolaris.org
-- 
Adam Leventhal, Solaris Kernel Development       http://blogs.sun.com/ahl