[ovs-dev] GRE TUNNEL

2012-04-02 Thread nicolas prochazka
Hello, since I'm using linux kernel 3.0.0 + openvswitch 1.4.0 in kernel module , my Gre tunnel does not work, Apr 2 14:11:46 positronic27900 ovs-vsctl: 1|vsctl|INFO|Called as ovs-vsctl -- add-port br0 GRE-10.98.98.25 tag=999 -- set interface GRE-10.98.98.25 type=gre options=local_ip=10.98.98.

Re: [ovs-dev] GRE TUNNEL

2012-04-02 Thread Kyle Mestery (kmestery)
On Apr 2, 2012, at 7:33 AM, nicolas prochazka wrote: > Hello, > since I'm using linux kernel 3.0.0 + openvswitch 1.4.0 in kernel module , > my Gre tunnel does not work, > > Apr 2 14:11:46 positronic27900 ovs-vsctl: 1|vsctl|INFO|Called as > ovs-vsctl -- add-port br0 GRE-10.98.98.25 tag=999

Re: [ovs-dev] GRE TUNNEL

2012-04-02 Thread Kyle Mestery (kmestery)
On Apr 2, 2012, at 7:41 AM, Kyle Mestery (kmestery) wrote: > On Apr 2, 2012, at 7:33 AM, nicolas prochazka wrote: >> Hello, >> since I'm using linux kernel 3.0.0 + openvswitch 1.4.0 in kernel module , >> my Gre tunnel does not work, >> >> Apr 2 14:11:46 positronic27900 ovs-vsctl: 1|vsctl|I

[ovs-dev] Openvswitch support - port tagging

2012-04-02 Thread Daniele Milani
I was looking at the release candidate of libvirt (0.9.11-rc2) and I have not been able to understand if the openvswitch support includes the port tagging support. So my question is: is it possible to create a tagged port? How? Otherwise, would it be useful to add this functionality? Thanks,

Re: [ovs-dev] Initial VXLAN Support in OVS

2012-04-02 Thread Kyle Mestery (kmestery)
On Mar 30, 2012, at 8:55 PM, Jesse Gross wrote: > On Thu, Mar 29, 2012 at 1:12 PM, Kyle Mestery (kmestery) > wrote: >> This patch builds on top of the VXLAN patch sent by Ben to the OVS list >> from October 2011. It is not complete, but I'd like to get some initial >> feedback on the approach here

Re: [ovs-dev] GRE TUNNEL

2012-04-02 Thread nicolas prochazka
hello again, with or without gre module kernel I obtain the same result. < > IP: GRE demultiplexer and <*> Open vSwitch Regards, Nicolas 2012/4/2 Kyle Mestery (kmestery) > On Apr 2, 2012, at 7:41 AM, Kyle Mestery (kmestery) wrote: > > On Apr 2, 2012, at 7:33 AM, nicolas prochazka wrote: >

Re: [ovs-dev] [PATCH] check-structs: Disallow uint_t because ovs_be should always be used.

2012-04-02 Thread Ben Pfaff
Thanks Simon, Ethan. I pushed this to master. On Mon, Apr 02, 2012 at 10:02:50AM +0900, Simon Horman wrote: > On Fri, Mar 30, 2012 at 09:24:06AM -0700, Ben Pfaff wrote: > > The header files that check-structs checks should only contain big-endian > > data, never native-endian data, so disallow ui

Re: [ovs-dev] Why updating ports upon ofproto_port_add?

2012-04-02 Thread Ben Pfaff
On Sat, Mar 31, 2012 at 04:37:41PM +0800, Min Chen wrote: > I am reading source code of Open vSwitch 1.4.0. I have a question on > updating ports immediately upon ofproto_port_add This is a strange question. What prompts the question, that is, why does it matter to you when we run it? > I search

Re: [ovs-dev] GRE TUNNEL

2012-04-02 Thread Justin Pettit
The "options" column is followed by a ":" not a "=". Also, you need to specify each option separately, so your options arguments would look more like this (not tested): options:local_ip=10.98.98.1 options:remote_ip=10.98.98.25 --Justin On Apr 2, 2012, at 5:33 AM, nicolas prochazka wr

Re: [ovs-dev] GRE TUNNEL

2012-04-02 Thread Ben Pfaff
Both "column=a=b,c=d" and "column:a=b column:c=d" should work and have the same effect, assuming the column is initially empty. This is documented in ovs-vsctl(8). On Mon, Apr 02, 2012 at 10:07:05AM -0700, Justin Pettit wrote: > The "options" column is followed by a ":" not a "=". Also, you need

Re: [ovs-dev] GRE TUNNEL

2012-04-02 Thread nicolas prochazka
I've tested with both ( column = and column: ) , definition is accepted in all cases but I get the same error. Before, I 'm using kernel 3.1.X with openvswitch module from branch 1.4 without this issues. Nicolas 2012/4/2 Ben Pfaff > Both "column=a=b,c=d" and "column:a=b column:c=d" should wor

Re: [ovs-dev] GRE TUNNEL

2012-04-02 Thread Justin Pettit
Oh, yeah. So I just tried the command on a 3.2.1 kernel with the tip of master, and it worked for me. Nicolas, maybe this is a silly question, but you're using the kernel module that came with the 1.4.0 release, right? You're not using the one that was sent for inclusion upstream, for example

[ovs-dev] (no subject)

2012-04-02 Thread Reliance Funding
Contact us for business and personal loans at 3%.Serious enquiries only.___ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev

Re: [ovs-dev] GRE TUNNEL

2012-04-02 Thread nicolas prochazka
No. I'm using openvswitch module in kernel 3.3.0 + openvswitch 1.4. branch for user land => not working Before I'm using openvswitch 1.4.x + include module on linux kernel 3.1.x => working But perhaps i do something wrong. Nicolas 2012/4/2 Justin Pettit > Oh, yeah. So I just tried the comman

Re: [ovs-dev] GRE TUNNEL

2012-04-02 Thread Justin Pettit
Oh, so you're using 3.3? Your original message said 3.0. The upstreamed kernel module doesn't have the built-in support for GRE tunnels. You should be able to use Linux's built-in GRE module (and treat it as a normal system port in OVS) or grab the latest OVS code from git and build the kernel

[ovs-dev] [PATCH] unixctl.py: Allow callers to manually set unixctl version.

2012-04-02 Thread Ethan Jackson
Some clients of unixctl.py may want to use a different version than the one supplied in ovs.version.VERSION. This patch supports an optional manual override of this value. Feature #10383. Signed-off-by: Ethan Jackson --- python/ovs/unixctl.py | 12 +++- 1 file changed, 7 insertions(+)

Re: [ovs-dev] GRE TUNNEL

2012-04-02 Thread nicolas prochazka
Ok sorry for that. and thanks for the explanation Regards, Nicolas 2012/4/2 Justin Pettit > Oh, so you're using 3.3? Your original message said 3.0. The upstreamed > kernel module doesn't have the built-in support for GRE tunnels. You should > be able to use Linux's built-in GRE module (and tre

Re: [ovs-dev] [PATCH] unixctl.py: Allow callers to manually set unixctl version.

2012-04-02 Thread Ben Pfaff
On Fri, Mar 30, 2012 at 11:47:52PM -0700, Ethan Jackson wrote: > Some clients of unixctl.py may want to use a different version than > the one supplied in ovs.version.VERSION. This patch supports an > optional manual override of this value. > > Feature #10383. > Signed-off-by: Ethan Jackson Cou

Re: [ovs-dev] [PATCH] unixctl.py: Allow callers to manually set unixctl version.

2012-04-02 Thread Ethan Jackson
I added this paragraph as a docstring to the create command: """Creates a new UnixctlServer which listens on a unixctl socket created at 'path'. If 'path' is None, the default path is chosen. 'version' contains the version of the server as reported by the unixctl ve

Re: [ovs-dev] [PATCH] ovs-test: Enhancments to the ovs-test tool

2012-04-02 Thread Ben Pfaff
On Fri, Mar 30, 2012 at 01:10:26PM -0700, Ansis Atteka wrote: > -Implemented support for ovs-test client, so that it could automatically > spawn an ovs-test server process from itself. This reduces the number of > commands the user have to type to get tests running. > -Automated creation of OVS bri

Re: [ovs-dev] [PATCH] ofp-util: Recognise wildcarded VLAN as NXM only.

2012-04-02 Thread Ethan Jackson
Is there any reason we can't simply ignore the CFI bit? Something like the following? I'm having trouble working out exactly how the CFI bit in OpenFlow matches interacts with the various versions and the classifier. Ethan --- lib/ofp-util.c |9 + 1 file changed, 9 insertions(+) d

Re: [ovs-dev] loopback tunnel not delivering

2012-04-02 Thread Jesse Gross
On Sun, Apr 1, 2012 at 5:45 PM, Dan Mihai Dumitriu wrote: > Hi all, > > I was wondering if anyone else had problems with gre, or capwap, tunnels > over the loopback interface.  (We're doing some automated tests which test > tunnels.)  What I'm seeing is that one packets gets through, and subsequen

Re: [ovs-dev] Openvswitch support - port tagging

2012-04-02 Thread Ansis Atteka
On Mon, Apr 2, 2012 at 5:56 AM, Daniele Milani wrote: > I was looking at the release candidate of libvirt (0.9.11-rc2) and I > have not been able to understand if the openvswitch support includes the > port tagging support. > If your question is "Whether it is possible to specify VLAN tag in Dom