On Mon, Jul 18, 2011 at 12:32:24PM -0700, Andrew Evans wrote:
> On Fri, Jul 15, 2011 at 3:12 PM, Ben Pfaff <[email protected]> wrote:
> > +# If openvswitch-switch is installed, and then BRCOMPAT is enabled in
> > +# /etc/default/openvswitch-switch, and then openvswitch-brcompat is
> > +# installed, start ovs-brcompatd.
> > +test -e /etc/default/openvswitch-switch && .
> > /etc/default/openvswitch-switch
> > +if test X"$1" = Xconfigure && \
> > + ? test X"$BRCOMPAT" = Xyes && \
> > + ? test -x /etc/init.d/openvswitch-switch && \
> > + ? test -e /var/run/openvswitch/ovs-vswitchd.pid && \
> > + ? test ! -e /var/run/openvswitch/ovs-brcompatd.pid; then
> > + ? ?invoke-rc.d openvswitch-switch start || exit $?
> > +fi
>
> I'm not sure I see the point in all these checks. Why not just start
> it regardless, since even in this scenario the init script will try to
> start daemons that are already running?
The scenarios I wanted to avoid were:
* ovs-vswitchd not running, admin installs
openvswitch-brcompat, postinst starts all the daemons.
* ovs-brcompatd not enabled, admin installs
openvswitch-brcompat, postinst wastes time just checking
that daemons are running and doesn't really do anything.
The former seems like an unnecessary annoyance to the admin. The
latter might not be worth worrying about, in which case I could drop
sourcing the default file and the $BRCOMPAT check. I'll make that
change, thanks.
_______________________________________________
dev mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/dev
