[ovs-dev] [lacp 01/10] vswitchd: Iface destruction should result in flow revalidation.

Ethan Jackson ethan at nicira.com
Tue Mar 15 23:46:41 UTC 2011


oops sorry.

Interfaces aren't deleted that often so I don't think the flush
matters that much from a performance perspective.

Ethan

On Tue, Mar 15, 2011 at 4:45 PM, Ben Pfaff <blp at nicira.com> wrote:
> OK.
>
> I don't understand your second sentence though.
>
> On Tue, Mar 15, 2011 at 04:43:11PM -0700, Ethan Jackson wrote:
>> I'm fine with ditching this patch.  Ifaces aren't deleted that option
>> so the flush is fine.
>>
>> Ethan
>>
>> On Tue, Mar 15, 2011 at 4:41 PM, Ben Pfaff <blp at nicira.com> wrote:
>> > I think that revalidation here is redundant with calling
>> > bridge_flush(). ?If you trace through that function's effects, you can
>> > see that it eventually causes every facet to be revalidated.
>> >
>> > I don't know why we do other explicit revalidation here. ?I guess that
>> > I didn't realize that it wasn't necessary.
>> >
>> > If we can actually get rid of the flush, that would be awesome, but
>> > that would take a more careful audit.
>> >
>



More information about the dev mailing list