On Mon, Jul 21, 2014 at 11:08 PM, Ben Pfaff <[email protected]> wrote:

> On Mon, Jul 21, 2014 at 04:55:57PM +0800, Kmindg G wrote:
> > When I set-controller to ovs in sandbox, I find that there're two
> > ovs-vswitchd threads.
> >
> >
> > ps  -L -p 19336
> >   PID   LWP TTY          TIME CMD
> > 19336 19336 ?        00:00:00 ovs-vswitchd
> > 19336 19368 ?        00:00:00 handler1
> > 19336 19369 ?        00:00:00 revalidator2
> > 19336 19370 ?        00:00:00 urcu3
> >
> > ovs-vsctl set-controller br0 tcp:3.3.3.3
> >
> > ps -L -p 19336
> >
> >   PID   LWP TTY          TIME CMD
> > 19336 19336 ?        00:00:00 ovs-vswitchd
> > 19336 19370 ?        00:00:00 urcu3
> > 19336 19615 ?        00:00:00 handler5
> > 19336 19616 ?        00:00:00 revalidator4
> > 19336 19627 ?        00:00:00 ovs-vswitchd
> >
> > ps -L -p 19336
> >   PID   LWP TTY          TIME CMD
> > 19336 19336 ?        00:00:00 ovs-vswitchd
> > 19336 19370 ?        00:00:00 urcu3
> > 19336 19615 ?        00:00:00 handler5
> > 19336 19616 ?        00:00:00 revalidator4
> > 19336 19712 ?        00:00:00 ovs-vswitchd
> >
> > 1. I don't change n_handler_thread or n_revalidator_thread, but handler
> and
> > revalidator threads change.
> > 2. thread id of the second ovs-vswitchd changes all the time.
> > Is this the expected behavior?
>
> This is probably glibc starting and later killing a thread to do
> asynchronous writes to the log file.  glibc doesn't give the thread a
> special name, so it inherits the process name.
>
> If you're concerned, you could turn off logging to a file (remove
> --log-file from the command line) to verify that this is the cause.
>

You are definitely right. Thanks for the hint.
_______________________________________________
discuss mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/discuss

Reply via email to