Re: [ovs-discuss] Connect VM and OVS more explanation

On Fri, Mar 13, 2015 at 06:43:29AM +0300, Amer wrote: > I have one VM1 as a host and another (VM2) has openvswitch installed in it. > > I have created a bridge br0 in the OVS(VM2) and assigned to it eth0,1,2 . I > attached eth0 in VM2 and VM1, both of them are virtualbox, to internal-network > Et

Re: [ovs-discuss] Connect VM and OVS more explanation

Hello, I think my previous message needs some clarification I have one VM1 as a host and another (VM2) has openvswitch installed in it. I have created a bridge br0 in the OVS(VM2) and assigned to it eth0,1,2 . I attached eth0 in VM2 and VM1, both of them are virtualbox, to internal-network Eth0

Re: [ovs-discuss] How does the kernel space and usespace receive netlink message

Please don't drop the mailing list. A couple of people have built prototypes that combine the two messages, but the performance difference was insignificant, so we did not commit it. On Fri, Mar 13, 2015 at 10:19:21AM +0800, Vinllen Chen wrote: > Hi, Ben, i have one more question: > Once the user

Re: [ovs-discuss] VxLAN kernel module.

On Thu, Mar 12, 2015 at 2:29 AM, Sudipto Biswas wrote: > Hi Pravin, Fan Du, > > Appreciate your help. > > To conclude, for VxLAN to work on Kernel 3.10 - do I need to explicitly load > the vxlan.ko and then re-build OVS to make use of it? > > I can see that CONFIG_VXLAN is set to 'm' > > Do I need

Re: [ovs-discuss] match/push/pop more than 3 MPLS labels

On Wed, Mar 11, 2015 at 5:01 AM, Stefano Salsano wrote: > Il 27/02/2015 00:52, Pravin Shelar ha scritto: >> >> On Thu, Feb 26, 2015 at 5:17 AM, Stefano Salsano wrote: >>> >>> >>> there is a limit of 3 MPLS labels that are supported by Open vSwitch >>> we need to raise it at 5 labels, to implement

Re: [ovs-discuss] OVS+NSH

On Mar 12, 2015, at 1:38 PM, Ben Pfaff mailto:b...@nicira.com>> wrote: [adding pritesh] On Thu, Mar 12, 2015 at 08:53:38PM +0100, christophe.fonta...@qosmos.com wrote: On March 21-22, I will participate to a hackathon hold before IETF 92. Within the featu

Re: [ovs-discuss] OVS+NSH

[adding pritesh] On Thu, Mar 12, 2015 at 08:53:38PM +0100, christophe.fonta...@qosmos.com wrote: > On March 21-22, I will participate to a hackathon hold before IETF 92. > Within the featured technologies, OVS is used by OpenDayLight : > - SFC-OVS Integration > - SFC/NSH > --> A test implement

[ovs-discuss] Connect VM and OVS

Hello, I have one VM1 as host and another has VM2 openvswitch installed in it. I have created a bridge br0 in the OVS and assigned to it eth0,1,2 as ports. I attached eth0 in VM2 and VM1, both of them are virtualbox, internal-network Eth0 in VM1 has IP 192.168.1.2 the br0 has IP 192.168.1.1 No o

[ovs-discuss] OVS+NSH

Hello, On March 21-22, I will participate to a hackathon hold before IETF 92. Within the featured technologies, OVS is used by OpenDayLight : - SFC-OVS Integration - SFC/NSH --> A test implementation is written in Python - SFC Classifier --> May involve OVS I saw that Priteshk's NSH patch

Re: [ovs-discuss] How does the kernel space and usespace receive netlink message

Please don't drop the mailing list. On Fri, Mar 13, 2015 at 12:04:45AM +0800, Vinllen Chen wrote: > Thankyou for your reply, Ben! There are two derivative question: > 1.Does the datapath also use multiple threads for receving netlink message ? The kernel always processes Netlink messages in the s

Re: [ovs-discuss] UDP packets not dropping

Hi justin Yes the iperf traffic is surely flowing through OVS.Also the configured flows for dropping udp traffic are matching but still the udp traffic is not dropping. On Mar 12, 2015 1:28 PM, "Justin Pettit" wrote: > > > On Mar 12, 2015, at 12:52 AM, abhishek jain > wrote: > > > > Hi justin >

Re: [ovs-discuss] UDP packets not dropping

Hi Justin I captured packets on wireshark with udp as filter and I was not able to obtain UDP packets with UDP traffic rule matching in OVS. Thanks On Thu, Mar 12, 2015 at 2:03 PM, abhishek jain wrote: > Hi justin > > Yes the iperf traffic is surely flowing through OVS.Also the configured > fl

Re: [ovs-discuss] How does the kernel space and usespace receive netlink message

On Wed, Mar 11, 2015 at 07:14:05PM +0800, Vinllen Chen wrote: > When i read the source code of ovs, i found the userspace receives > netlink message by epoll. But in the explanition of dpif.h, datapath starts > many threads to receiving netlink. > Could anybody give me a definite answer tha

Re: [ovs-discuss] upstream Extending netdev-* port config options

Thank you Ben! I have discovered this in Ryu vsctl.py just now too :) On 3/12/15, 10:01 AM, "Ben Pfaff" wrote: >On Thu, Mar 12, 2015 at 08:50:36AM -0500, David Evans wrote: >> I have some fun extensions to things like netdev-dpdk and netdev-vport. >> I can control them using ovs-vsctl by adding

Re: [ovs-discuss] Proxying using openvswitch

On Thu, Mar 12, 2015 at 03:00:46PM +, Jason Kölker wrote: > You can use `action=in_port` to send it back from the same port it > came in as. However I think that this always uses the userspace path, > and not a datapath flow. No, there's no special case here. __

Re: [ovs-discuss] Proxying using openvswitch

The FAQ says: ### Q: I added a flow to send packets out the ingress port, like this: ovs-ofctl add-flow br0 in_port=2,actions=2 but OVS drops the packets instead. A: Yes, OpenFlow requires a switch to ignore attempts to send a packet out its ingress port. The rationale is that dro

Re: [ovs-discuss] upstream Extending netdev-* port config options

On Thu, Mar 12, 2015 at 08:50:36AM -0500, David Evans wrote: > I have some fun extensions to things like netdev-dpdk and netdev-vport. > I can control them using ovs-vsctl by adding ³options:foo=on ³ type of > fields. > > Question: What else upstream do I need to touch to be able to do likewise >

Re: [ovs-discuss] Proxying using openvswitch

On Thu, Mar 12, 2015 at 11:59 AM, Emma Anderson wrote: > Hi all, > > Can open vswitch acts like: receive a packet from eth0, substitute IP/mac, > and sends it back again on eth0? > > I found that when a packet is received from eth0, although the flow is > triggered, it looks that packet is dropped

Re: [ovs-discuss] Proxying using openvswitch

The issue here—if I understand correctly—is that OpenFlow doesn't allow traffic to be forwarded out the same port in which it was received. To make this work, you'd have to receive traffic on eth0 and send it out eth1 (for example). Happy to be corrected if I am mistaken. -- Scott Sent from m

[ovs-discuss] Proxying using openvswitch

Hi all, Can open vswitch acts like: receive a packet from eth0, substitute IP/mac, and sends it back again on eth0? I found that when a packet is received from eth0, although the flow is triggered, it looks that packet is dropped silently and there is no trace in logs (if I am looking at the ri

Re: [ovs-discuss] Linear topology in openveswitch

Thank you I have fixed it Eng Amer Alghadhban COE SANS-GCFW CEH, SCNP, CCNA From: amer7...@hotmail.com To: discuss@openvswitch.org Date: Wed, 11 Mar 2015 18:35:28 + Subject: Re: [ovs-discuss] Linear topology in openveswitch Unfortunately, it doesn't work I got this error, when I

Re: [ovs-discuss] VxLAN kernel module.

To correct myself, will unloading and re-loading the openvswitch module reference the vxlan.ko module automatically, if it's loaded? Thanks, Sudipto On Thursday 12 March 2015 02:59 PM, Sudipto Biswas wrote: Hi Pravin, Fan Du, Appreciate your help. To conclude, for VxLAN to work on Kernel 3.10

Re: [ovs-discuss] VxLAN kernel module.

Hi Pravin, Fan Du, Appreciate your help. To conclude, for VxLAN to work on Kernel 3.10 - do I need to explicitly load the vxlan.ko and then re-build OVS to make use of it? I can see that CONFIG_VXLAN is set to 'm' Do I need to re-build OVS to use this module? (after i have used modprobe to

Re: [ovs-discuss] UDP packets not dropping

> On Mar 12, 2015, at 12:35 AM, abhishek jain wrote: > > Yes I want to drop the whole UDP traffic.How Is it possible with OVS? Isn't that what's happening, since the packet counts are going up for that flow? > > ovs-ofctl dump-flows br0 > > NXST_FLOW reply (xid=0x4): > > cookie=0x0, duration=

Re: [ovs-discuss] UDP packets not dropping

> On Mar 12, 2015, at 12:29 AM, abhishek jain wrote: > > Hi Justin.. > > Sorry for the typo error.I'm settig the priority of the UDP higher than the > other rules.The rule to drop UDP packet is matching but I'm not able to drop > the Iperf UDP traffic. Below is the command I'm using.. > > o

Re: [ovs-discuss] UDP packets not dropping

Hi Justin.. Sorry for the typo error.I'm settig the priority of the UDP higher than the other rules.The rule to drop UDP packet is matching but I'm not able to drop the Iperf UDP traffic. Below is the command I'm using.. ovs-ofctl add-flow br0 dl_type=0x0800,nw_proto=17,priority=18,actions=drop

Re: [ovs-discuss] UDP packets not dropping

Hi Justin We have changed the priority of other flows also. ovs-ofctl dump-flows br0 NXST_FLOW reply (xid=0x4): cookie=0x0, duration=73.296s, table=0, n_packets=0, n_bytes=0, idle_age=73, priority=15,udp,tp_src=5001 actions=drop cookie=0x0, duration=162.284s, table=0, n_packets=2386, n_bytes=8

Re: [ovs-discuss] UDP packets not dropping

> On Mar 11, 2015, at 11:54 PM, Navneet Singh > wrote: > > Hi Justin, > > I changed the command as per your recommendation to > ovs-ofctl add-flow br0 > dl_type=0x0800,nw_proto=17,priority=15,udp_src=5001,actions=drop > > The flows are not matching when I'm sending traffic through iperf udp