[ovs-discuss] [CudaMailTagged] Traffic can't get through OVS

[Pavel Gluschak]

sdavis <steven.davis at ...> writes:

> 
> Greetings!
> 
> I’ve installed OVS with some NSH patches using these instructions:
> https://github.com/yyang13/ovs_nsh_patches/blob/master/README.md. I ended
up building an rpm
> from the resulting package and installed from that. Here’s version output:
> 
> [root <at> prv-10-11-ksupp user]# ovs-vsctl -V
> ovs-vsctl (Open vSwitch) 2.5.90
> Compiled May  6 2016 11:59:34
> DB Schema 7.12.1
> 
> My configuration is very simple as seen here:
> 
> [root <at> prv-10-11-ksupp user]# ovs-vsctl show
> 4f8721e2-43a2-4cf0-b43c-bb74b51904ac
>     Bridge "br0"
>         Port "br0"
>             Interface "br0"
>                 type: internal
>         Port "vnet0"
>             Interface "vnet0"
>         Port "vnet1"
>             Interface "vnet1"
>     ovs_version: “2.5.90"
> 
> and, at this point, I just want traffic to flow between 2 vm’s on the host
itself, hence this flow:
> 
> [root <at> prv-10-11-ksupp user]# ovs-ofctl dump-flows br0
> NXST_FLOW reply (xid=0x4):
>  cookie=0x0, duration=63144.743s, table=0, n_packets=0, n_bytes=0,
idle_age=63144, priority=0 actions=NORMAL
> 
> Super simple install it would appear, yet I can’t get vm1 to talk to vm2
through br0… I switch it out for a
> linux bridge and it works fine, but ovs is a no go… Here are the log
entries that are consistently filling
> my logs during a test ping (that fails) between the vm’s:
> 
> May 10 11:21:09 prv-10-11-ksupp ovs-vswitchd:
ovs|01944|odp_util(handler19)|ERR|Dropped 5 log
> messages in last 5 seconds (most recently, 1 seconds ago) due to excessive
rate
> May 10 11:21:09 prv-10-11-ksupp ovs-vswitchd:
ovs|01945|odp_util(handler19)|ERR|attribute
> tcp_flags has length 4 but should have length 2
> May 10 11:21:15 prv-10-11-ksupp ovs-vswitchd:
ovs|01946|odp_util(handler19)|ERR|Dropped 5 log
> messages in last 5 seconds (most recently, 1 seconds ago) due to excessive
rate
> May 10 11:21:15 prv-10-11-ksupp ovs-vswitchd:
ovs|01947|odp_util(handler19)|ERR|attribute
> tcp_flags has length 4 but should have length 2
> May 10 11:21:21 prv-10-11-ksupp ovs-vswitchd:
ovs|01948|odp_util(handler19)|ERR|Dropped 5 log
> messages in last 5 seconds (most recently, 1 seconds ago) due to excessive
rate
> May 10 11:21:21 prv-10-11-ksupp ovs-vswitchd:
ovs|01949|odp_util(handler19)|ERR|attribute
> tcp_flags has length 4 but should have length 2
> May 10 11:21:27 prv-10-11-ksupp ovs-vswitchd:
ovs|01950|odp_util(handler19)|ERR|Dropped 8 log
> messages in last 5 seconds (most recently, 1 seconds ago) due to excessive
rate
> May 10 11:21:27 prv-10-11-ksupp ovs-vswitchd:
ovs|01951|odp_util(handler19)|ERR|attribute
> tcp_flags has length 4 but should have length 2
> 
> When I stop a ping, the logs stop. Also here’s the output of ova-dbctl show:
> 
> [root <at> prv-10-11-ksupp user]# ovs-dpctl show
> system <at> ovs-system:
> 	lookups: hit:0 missed:3819 lost:0
> 	flows: 0
> 	masks: hit:0 total:0 hit/pkt:0.00
> 	port 0: ovs-system (internal)
> 	port 1: br0 (internal)
> 	port 2: vnet0
> 	port 3: vnet1
> 
> Any idea’s why this might be happening? tcpdumping on the on the vnet
shows the icmp traffic leaving vm1
> but I don’t see anything if I tcpdump on br0…
> 
> Thanks in advance!
> _______________________________________________
> discuss mailing list
> discuss <at> openvswitch.org
> http://openvswitch.org/mailman/listinfo/discuss
> 

OVS with NSH patches does not support kernel dataplane.
You need to make bridge work in userspace mode by:
ovs-vsctl set bridge br0 datapath_type=netdev