[ovs-dev] [PATCH V9 1/3] ovn-controller-vtep: Add vtep module.

[Justin Pettit]

> On Sep 13, 2015, at 11:35 AM, ALeX Wang <ee07b291 at gmail.com> wrote:
> 
>> On 12 September 2015 at 16:43, Justin Pettit <jpettit at nicira.com> wrote:
>> 
>> > On Aug 27, 2015, at 11:21 PM, Alex Wang <ee07b291 at gmail.com> wrote:
>> >
>> 
>> > +/* Since we do not own any vtep logical switch, just sets their tunnel key
>> > + * to 0. */
>> > +static bool
>> > +vtep_lswitch_cleanup(struct ovsdb_idl *vtep_idl)
>> > +{
>> > +    const struct vteprec_logical_switch *vtep_ls;
>> > +    int64_t tnl_key = 0;
>> > +    bool done = true;
>> > +
>> > +    VTEPREC_LOGICAL_SWITCH_FOR_EACH (vtep_ls, vtep_idl) {
>> > +        if (vtep_ls->n_tunnel_key != 1
>> > +            || vtep_ls->tunnel_key[0] != tnl_key) {
>> > +            vteprec_logical_switch_set_tunnel_key(vtep_ls, &tnl_key, 1);
>> 
>> It looks like the tunnel key is optional.  This code will set a tunnel key even if one wasn't previously set.  Is this intentional?  I was just surprised to see this in a cleanup function.
> 
> I'm also unclear about if this is necessary.  The reason I wanted to reset it, is
> because the controller sets it.  And since there is not a good way to mark which
> tunnel key is set by controller, I just reset them all.
> 
> Maybe we can just remove this function?

I like that it cleans up.  Since 0 is a valid value, I wonder if the better behavior is to actually remove the tunnel key value entirely.

--Justin