Hi OVS wizards,
Further to my query below,
if in my compose action i turn off 'was_mpls' (in my case
'was_encapsulated')
The bridge will send the packet across the patch port.
which leads me to ask:
    is there a way to make if ctx->xbridge->support.odp.recirc false for
one bridge and not another via ovs-vsctl or ovs-ofctl command?
Admittedly i would much prefer to NOT need to do this, but instead to
always ignore the recirc if an output action is on the rule.

Regards,
Dave.

On Tue, May 31, 2016 at 9:26 AM, David Evans <[email protected]>
wrote:

> Hi OVS wizards,
> i have
> a custom action with an output
> and
> a default rule to drop packets.
>
>  cookie=0x5000000000003, duration=320317.501s, table=0, n_packets=4,
> n_bytes=7984, send_flow_rem priority=99,gre,in_port=1,nw_dst=7.7.7.7
> actions=strip_gre,output:2
>  cookie=0x4000000000004, duration=320319.766s, table=0, n_packets=4,
> n_bytes=7832, send_flow_rem priority=0 actions=drop
>
> now my custom action appears to work, and the output is a patch to another
> bridge, but the packet never leaves this bridge, and the drop action
> catches the packet instead (as you can see above - 4 packets hit the drop
> rule (that are a little shorter than the 4 packets which hit the gre_strip,
> output rule.))
>
> So i have a recirculation problem?
> My custom action checks recirculation because i have stripped GRE (like
> you do with MPLS - so i changed the name to ENCAPSULATION), but I want the
> output action to apply and not to fall through to the next rule.
>
> Am i doing it wrong? This used to work with ovs 2.3.2ish and now doesn't
> seem to do the same thing in ovs 2.5.0
> Has a recirculation capability changed with regard to output actions?
>
> do_xlate_actions()....
>
>         case OFPACT_STRIP_GRE:
>             CHECK_ENCAPSULATION_RECIRCULATION();
>                 compose_strip_gre_action(ctx);
>             break;
>
> Regards,
> Dave.
>
>
_______________________________________________
discuss mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/discuss

Reply via email to