Hi Ben,

Ok, i get it. So, vswitchd tells the kernel to delete idle flows, but
vswitchd itself will still have the flow information right ? i.e. even
though the flow gets deleted from the kernel, next time when that flow comes
to the switch, vswitchd will be able to handle this flow(by adding it t teh
kernel again), without me having to again configure the flow using ovs-ofctl
add flow..?

Thanks,
AIsh

On Wed, Sep 21, 2011 at 12:58 PM, Ben Pfaff <[email protected]> wrote:

> On Wed, Sep 21, 2011 at 12:51:27PM -0700, Aishwarya wrote:
> > I did not quite understand your reply to the first question. what does
> the
> > expression "openvswitch_mod.ko does no flow expiration" mean then? Also,
> you
> > mentioned "userspace deletes flows that fall idle" : you mean vswitchd
> > deletes them from the database?
>
> The kernel doesn't do flow expiration.  ovs-vswitchd tells the kernel
> when to delete idle flows.
>
> Flows aren't kept in the database.
>
> > For my use case, what I am doing is
> > configuring all required flows when openvswitch starts up, and i require
> > them to exist whenever necessary unless the vswitchd process actually
> dies.
> > Is there a way to do this? (I need flows to always exist in vswitchd).
>
> No.
>
> > If this is not possible, how do we know when to re-push the flow rules?
>
> ovs-vswitchd does it.
>
> If you need bridging to work without help from a userspace process,
> then you can't use OVS.  Period.
>
> > A side question: Is flow information actually stored in the database, or
> > does the database hold only port/bridge information. I do not see any
> flow
> > table in the database document description.
>
> Flows aren't kept in the database.
>
_______________________________________________
discuss mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/discuss

Reply via email to