After some further testing, the whole hypervisor with unrelated VMs becomes
unresponsive.
OVS is running inside a Debian Stretch VM, not on the hypervisor.
As soon as I add these rules, the packetloss to the server reaches about
40% - even from hosts that have nothing to do with this bridge:
#
Seems like it works :-)
I used the wrong id for the in_port directive and did not know there is an
action=LOCAL.
Just get started to learn OVS.
The mesh network needs to settle down a bit but it looks like it works!
Thank you very much!
Kind regards
Kevin
Am Di., 14. Mai 2019 um 08:53 Uhr
Ok, there already is "br0" as int interface, didn't read correctly.
Your command is missing some pieces, can you check again?
Kind regards
Kevin
Am Di., 14. Mai 2019 um 08:32 Uhr schrieb Kevin Olbrich :
> Hi Matthias,
>
> do I need to create an "int" port for this?
> Currently I bind an IP
Hi Matthias,
do I need to create an "int" port for this?
Currently I bind an IP directly to br0.
Thank you!
Kind regards
Kevin
Am Di., 14. Mai 2019 um 08:00 Uhr schrieb Matthias May via discuss <
ovs-discuss@openvswitch.org>:
> On 14/05/2019 07:26, Kevin Olbrich wrote:
> > Hi!
> >
> > I've
On 14/05/2019 07:26, Kevin Olbrich wrote:
> Hi!
>
> I've got an OVS that has a bridge "br0" and has about 100x L2TP tunnels.
> These tunnels run batman-adv, a mesh protocol for L2 routing over L3.
>
> For efficient routing, only nodes that are in the same building are allowed
> to see each
Hi!
I've got an OVS that has a bridge "br0" and has about 100x L2TP tunnels.
These tunnels run batman-adv, a mesh protocol for L2 routing over L3.
For efficient routing, only nodes that are in the same building are allowed
to see each other.
To filter out traffic between the ports, I used