On Tue, Feb 02, 2016 at 07:58:32AM +0300, Ilya Maximets wrote:
> On 01.02.2016 21:31, Ben Pfaff wrote:
> > On Mon, Feb 01, 2016 at 10:24:29AM +0300, Ilya Maximets wrote:
> >> On 29.01.2016 22:02, Guru Shetty wrote:
> >>> This reverts commit 9b5422a98f817b9f2a1f8224cab7e1a8d0bbba1f.
> >>>
> >>> Reas
On 29.01.2016 04:44, Daniele Di Proietto wrote:
> I'm happy with the patch, I just wanted to test it a little more
> with vhost queue.
>
> If everything looks good I'll merge it tomorrow
Is there some issues with vhost?
Best regards, Ilya Maximets.
___
On 01.02.2016 21:31, Ben Pfaff wrote:
> On Mon, Feb 01, 2016 at 10:24:29AM +0300, Ilya Maximets wrote:
>> On 29.01.2016 22:02, Guru Shetty wrote:
>>> This reverts commit 9b5422a98f817b9f2a1f8224cab7e1a8d0bbba1f.
>>>
>>> Reason:
>>> Calling 'ovs-appctl exit' on ovs-vswitchd will cause ovs-vswitchd
>
Hi all,
I have noticed that OVS support multiple tunneling protocols, such as, GRE,
VXLAN, etc. I am not sure if the MPLS has been also supported now (could not
find it ...). Since the openflow protocol has support the corresponding actions
of MPLS, such as, push_mpls, pop_mpls. I have try to t
On 01/02/2016 12:30, "Chandran, Sugesh" wrote:
>Hi Daniele,
>
>Thank you for sending out the patch.
>The proposal looks fine for me and also I verified that the performance
>is restored with the patch.
Thanks!
>
>One suggestion is , Do we really have to pass two parameters to the
>"dp_netdev_i
When a group of packets arrives from a port, we loop through them to
initialize metadata and then we loop through them again to extract the
flow and perform the exact match classification.
This commit combines the two loops into one, and initializes packet->md
in emc_processing() to improve perfor
On 30/01/2016 17:30, "Andy Zhou" wrote:
>
>
>On Thu, Jan 28, 2016 at 5:56 PM, Daniele Di Proietto
> wrote:
>
>When a group of packets arrives from a port, we loop through them to
>initialize metadata and them we loop through them again to extract the
>flow and perform the exact match classifica
Thanks for the patch!
Acked-by: Daniele Di Proietto
On 01/02/2016 12:46, "Andy Zhou" wrote:
>Commit d262ac2c60ce1da7b477737f70e8efd38b32502d introduced a slight
>performance drop for the fast path, where every packets hits the
>emc cache. This patch removes that performance drop by only reloa
On 01/02/2016 12:46, "Andy Zhou" wrote:
>For the machines I have access to, Reloading the same pointer from
>memory seems to inhibit complier optimization somewhat.
>
>In emc_processing(), using a single packet pointer, instead reloading
>it from memory with packets[i], improves performance by
On 29 January 2016 at 12:10, Serge Hallyn wrote:
> Quoting Christian Ehrhardt (christian.ehrha...@canonical.com):
> > On Wed, Jan 27, 2016 at 8:26 PM, Ansis Atteka
> wrote:
> >
> > >
> > >
> > > On 27 January 2016 at 02:30, Christian Ehrhardt <
> > > christian.ehrha...@canonical.com> wrote:
> >
On Mon, Feb 01, 2016 at 10:51:58PM +0100, Dimitri John Ledkov wrote:
> Build log for powerpc platform:
>
> https://launchpadlibrarian.net/235964133/buildlog_ubuntu-xenial-powerpc.openvswitch_2.5.0~git20160129.46a88d9-0ubuntu2.1_BUILDING.txt.gz
>
> Ctrl-F for "testsuite: command line was:" to jump
Thanks, applied to master, branch-2.5, and branch-2.4.
On Mon, Feb 01, 2016 at 10:16:41PM +0100, Dimitri John Ledkov wrote:
> lgtm. +1
>
> I have came to the same conclusion starring at this code at FOSDEM,
> but didn't fix all the tests.
>
>
>
> On 1 February 2016 at 21:55, Ben Pfaff wrote:
On 1 February 2016 at 22:51, Dimitri John Ledkov wrote:
> Build log for powerpc platform:
>
> https://launchpadlibrarian.net/235964133/buildlog_ubuntu-xenial-powerpc.openvswitch_2.5.0~git20160129.46a88d9-0ubuntu2.1_BUILDING.txt.gz
>
> Ctrl-F for "testsuite: command line was:" to jump to the "cat
>
Build log for powerpc platform:
https://launchpadlibrarian.net/235964133/buildlog_ubuntu-xenial-powerpc.openvswitch_2.5.0~git20160129.46a88d9-0ubuntu2.1_BUILDING.txt.gz
Ctrl-F for "testsuite: command line was:" to jump to the "cat
tests/testuite.log" output.
There are now 22 failures:
Subject:
Hello,
The reasoning sounds good, but i wouldn't understand this just by
looking at the patch.
Let me test build these both patches, and I'll let you know the results soon.
Regards,
Dimitri.
On 1 February 2016 at 21:55, Ben Pfaff wrote:
> struct flow has a union for the in_port field, and the
Hi Chris
Thank you for the clarification. You have some valuable arguments!
I think that when a person's involvement in a project is indeed based on
someone's work assignment, that it should be properly and openly disclosed to
the OPEN community in one way or another. Sounds, to me, like part
lgtm. +1
I have came to the same conclusion starring at this code at FOSDEM,
but didn't fix all the tests.
On 1 February 2016 at 21:55, Ben Pfaff wrote:
> ip_frag_off is an ovs_be16 so it must be converted between host and
> network byte order for parsing and formatting.
>
> Reported-by: Dimit
We had a bug filed against the OpenStack+OVN integration
(networking-ovn) that Neutron security group changes are not applied to
existing connections. The existing OVS integration in Neutron does this
by deleting conntrack state entries by running the conntrack tool from a
Python agent running on
On Fri, Jan 29, 2016 at 05:26:47PM -0800, Jarno Rajahalme wrote:
>
> > On Jan 29, 2016, at 5:09 PM, Ben Pfaff wrote:
> >
> > On Fri, Jan 29, 2016 at 03:38:12PM -0800, Jarno Rajahalme wrote:
> >> Handle implicit recirculation explicitly for each action type, so that
> >> it is easier to follow wh
On Mon, Feb 1, 2016 at 11:15 AM, Ben Pfaff wrote:
> On Fri, Jan 29, 2016 at 07:14:35PM -0800, Andy Zhou wrote:
>> This is essentially the same patch as the original patched posted at:
>> http://openvswitch.org/pipermail/dev/2016-January/064971.html
>>
>> The original commit message:
>> Before
ip_frag_off is an ovs_be16 so it must be converted between host and
network byte order for parsing and formatting.
Reported-by: Dimitri John Ledkov
Reported-at: http://openvswitch.org/pipermail/discuss/2016-January/020072.html
Signed-off-by: Ben Pfaff
---
lib/odp-util.c | 6 --
t
struct flow has a union for the in_port field, and the two different
members of the union have different sizes: ofp_port_t is 16 bits,
odp_port_t is 32 bits. On little-endian machines this doesn't matter much,
but on big-endian machines it's important to distinguish between them
because a small nu
This patch provides the modifications required in netdev-dpdk.c and
vswitch.xml to allow for a DPDK user space QoS algorithm.
This patch adds a QoS configuration structure for netdev-dpdk and
expected QoS operations 'dpdk_qos_ops'. Various helper functions
are also supplied.
Also included are the
Commit d262ac2c60ce1da7b477737f70e8efd38b32502d introduced a slight
performance drop for the fast path, where every packets hits the
emc cache. This patch removes that performance drop by only reloading
the key pointer on emc cache miss.
Sgned-off-by: Andy Zhou
---
lib/dpif-netdev.c | 7 ---
For the machines I have access to, Reloading the same pointer from
memory seems to inhibit complier optimization somewhat.
In emc_processing(), using a single packet pointer, instead reloading
it from memory with packets[i], improves performance by 0.3 Mpps (tested
with 10G NIC pushing 64 byte pac
Hi Daniele,
Thank you for sending out the patch.
The proposal looks fine for me and also I verified that the performance is
restored with the patch.
One suggestion is , Do we really have to pass two parameters to the
"dp_netdev_input" to init and validate the metadata.?
Can we use the pointer
On 02/01/2016 02:52 PM, Saurabh Shrivastava (सौरभ श्रीवास्तव) wrote:
> How about reusing Openstack Congress which is already out there and also
> implements something like nlog?
One simple reason is that Congress is written in Python, while OVN is
written in C.
--
Russell Bryant
How about reusing Openstack Congress which is already out there and also
implements something like nlog?
--
Saurabh (सौरभ)
On Mon, Feb 1, 2016 at 11:12 AM, Ben Pfaff wrote:
> On Mon, Feb 01, 2016 at 10:12:49AM -0500, Russell Bryant wrote:
> > On 02/01/2016 05:03 AM, Yusheng Wang wrote:
> > > W
Hi Antonio,
Thanks for investigating and sharing these results.
You're right, this patch changes slightly the assembly output in
miniflow_extract() and introduces a performance regression.
On 01/02/2016 03:13, "Fischetti, Antonio"
wrote:
>Hi, I was running some Regression Test to check perfor
Numan Siddique wrote on 01/02/2016 09:23:43 AM:
>
> On Mon, Feb 1, 2016 at 12:33 PM, Liran Schour wrote:
> Python idl works now with "monitor_cond" method. Add test
> for backward compatibility with old "monitor" method.
>
> Signed-off-by: Liran Schour
>
> ---
> v2->v3:
> *cond_update() receiv
On Fri, Jan 29, 2016 at 07:14:35PM -0800, Andy Zhou wrote:
> This is essentially the same patch as the original patched posted at:
> http://openvswitch.org/pipermail/dev/2016-January/064971.html
>
> The original commit message:
> Before this commit, emc_processing() copied a netdev_flow_key if
On Mon, Feb 01, 2016 at 10:12:49AM -0500, Russell Bryant wrote:
> On 02/01/2016 05:03 AM, Yusheng Wang wrote:
> > We are trying to propose a new computation model for the ovn-northd
> > application. Currently, ovn-northd computes all logical data path
> > flows for every change. This full computati
On 02/01/2016 12:21 PM, Ben Pfaff wrote:
> On Mon, Feb 01, 2016 at 10:01:23AM -0500, Russell Bryant wrote:
>> ovn-northd set some custom log level defaults, which I believe were
>> copied from ovs-vsctl. Other daemons don't set this. The difference in
>> behavior in ovn-northd vs other daemons ha
On 02/01/2016 01:04 PM, Chris Wright wrote:
> Hey Uri,
>
> * Elzur, Uri (uri.el...@intel.com) wrote:
>> I find it strange and misleading for full time employees to use aliases on
>> OPEN SOURCE mailing lists. Is the code OPEN but affiliation is not? Does
>> anyone work pro bono?
Thanks for shar
On Mon, Feb 01, 2016 at 10:24:29AM +0300, Ilya Maximets wrote:
> On 29.01.2016 22:02, Guru Shetty wrote:
> > This reverts commit 9b5422a98f817b9f2a1f8224cab7e1a8d0bbba1f.
> >
> > Reason:
> > Calling 'ovs-appctl exit' on ovs-vswitchd will cause ovs-vswitchd
> > to destroy all the ports from the dat
This is a proposal regarding how Floating IP will be supported in
OpenStack using OVN.
The complete proposal can also be found at
https://review.openstack.org/#/c/274274/.
The additions proposed to the OVN Northbound Schema and Southbound Flows
are outlined here as well. The proposal assumes t
On Mon, Feb 01, 2016 at 08:05:33AM -0800, Guru Shetty wrote:
> On 31 January 2016 at 23:24, Ilya Maximets wrote:
>
> > On 29.01.2016 22:02, Guru Shetty wrote:
> > > This reverts commit 9b5422a98f817b9f2a1f8224cab7e1a8d0bbba1f.
> > >
> > > Reason:
> > > Calling 'ovs-appctl exit' on ovs-vswitchd wi
Hey Uri,
* Elzur, Uri (uri.el...@intel.com) wrote:
> I find it strange and misleading for full time employees to use aliases on
> OPEN SOURCE mailing lists. Is the code OPEN but affiliation is not? Does
> anyone work pro bono?
This is actually a pretty normal practice for open source developers
On Thu, Jan 28, 2016 at 01:14:59AM +, Daniele Di Proietto wrote:
>
>
> On 26/01/2016 08:23, "Ben Pfaff" wrote:
>
> >On Tue, Jan 26, 2016 at 03:28:47AM +, Daniele Di Proietto wrote:
> >>
> >>
> >> On 25/01/2016 18:20, "Andy Zhou" wrote:
> >>
> >> >On Sun, Jan 24, 2016 at 8:32 AM, Ben
On Mon, Feb 01, 2016 at 01:50:42PM +0300, Ilya Maximets wrote:
> On 12.01.2016 19:27, Ben Pfaff wrote:
> > On Tue, Jan 12, 2016 at 02:15:39PM +0300, Ilya Maximets wrote:
> >> On arm/arm64:
> >>gcc: error: unrecognized command line option '-mssse3'
> >>
> >> Signed-off-by: Ilya Maximets
> >
>
On Mon, Feb 01, 2016 at 10:01:23AM -0500, Russell Bryant wrote:
> ovn-northd set some custom log level defaults, which I believe were
> copied from ovs-vsctl. Other daemons don't set this. The difference in
> behavior in ovn-northd vs other daemons has caused some confusion during
> OpenStack+OVN
On Mon, Feb 01, 2016 at 04:40:44PM -0800, Quentin Monnet wrote:
> Proposal: add an option to ovs-ofctl utility so as to obtain colorized output
> in tty, for easier reading. Currently, only the dump-flows command supports
> colors.
>
> A new `--color` option has been added to ovs-ofctl so as to in
Hi Zoltan. Thanks for your mail - responses inline below.
Cheers,
Mark
>
>Hi,
>
>I have a related question: is it a possible alternative with the current
>codebase to use KNI for configuring the MTU?
This would (as far as I'm aware) require the implementation of a new KNI netdev
class, along wi
On Tue, Jan 26, 2016 at 10:09:28PM -0800, Jesse Gross wrote:
> On Tue, Jan 19, 2016 at 10:15 PM, Simon Horman
> wrote:
> > Add support for layer 3 GRE vports (non-tap aka non-VTEP).
> >
> > This makes use of a separate vport type for GRE, rather than a new mode for
> > the existing (tap/VTEP) GRE
On 31 January 2016 at 23:24, Ilya Maximets wrote:
> On 29.01.2016 22:02, Guru Shetty wrote:
> > This reverts commit 9b5422a98f817b9f2a1f8224cab7e1a8d0bbba1f.
> >
> > Reason:
> > Calling 'ovs-appctl exit' on ovs-vswitchd will cause ovs-vswitchd
> > to destroy all the ports from the datapath. This
Hi,
I have a related question: is it a possible alternative with the current
codebase to use KNI for configuring the MTU?
I never actually tried KNI, does it have a performance penalty to just
have it?
The reason I'm asking these questions because OVS normally doesn't
configure the MTU, but it
Proposal: add an option to ovs-ofctl utility so as to obtain colorized output
in tty, for easier reading. Currently, only the dump-flows command supports
colors.
A new `--color` option has been added to ovs-ofctl so as to indicate
whether color markers should be used or not. It can be set to `alwa
On 02/01/2016 05:03 AM, Yusheng Wang wrote:
> We are trying to propose a new computation model for the ovn-northd
> application. Currently, ovn-northd computes all logical data path
> flows for every change. This full computation model is simple and
> helped verifying the concept of logical data pa
On Mon, Feb 1, 2016 at 9:01 AM, Russell Bryant wrote:
> ovn-northd set some custom log level defaults, which I believe were
> copied from ovs-vsctl. Other daemons don't set this. The difference in
> behavior in ovn-northd vs other daemons has caused some confusion during
> OpenStack+OVN developm
ovn-northd set some custom log level defaults, which I believe were
copied from ovs-vsctl. Other daemons don't set this. The difference in
behavior in ovn-northd vs other daemons has caused some confusion during
OpenStack+OVN development and testing, so make it consistent.
Reported-by: Ryan Moat
Olen j�sen Julie Leach, olen lahjoittaamiljoona dollaria (7) seitsem�n
onnekkaisiinjackpot voittaa, ja olet yksi onnekas voittajaonnittelut!
vain varmistaa, tavoittaa sinut suoraank�yt�n google K��nt�j� kirjoittaa voit
omallakielell�si yhteytt� minuun lis�tietojavaatimuksen. , S�hk�postitse
minu
I'm sharing some figures I got with perf tool.
conntrack_execute() takes 42%, which is about 3 times than the 15% from
emc_processing().
Below some details.
I did
git clone https://github.com/ddiproietto/ovs.git
git checkout userconntrack_20151115
I setup the following flows
sudo $OVS_DIR/util
Hi, I was running some Regression Test to check performance in a
Phy-2-Phy test, ie
ovs-ofctl add-flow br0 in_port=1,action=output:2
After applying patches #1 and then #2 the performance
is still ok.
Instead, after I apply this patch #3 the throughput goes from 12.1
down to 11.6 Mpps.
I found ou
As this Regression Test issue is specific for Patch #3/12
I'll continue this topic on the post for [Patch 03/12] at
http://openvswitch.org/pipermail/dev/2015-November/062230.html
> -Original Message-
> From: dev [mailto:dev-boun...@openvswitch.org] On Behalf Of Fischetti,
> Antonio
> Sent
On 12.01.2016 19:27, Ben Pfaff wrote:
> On Tue, Jan 12, 2016 at 02:15:39PM +0300, Ilya Maximets wrote:
>> On arm/arm64:
>> gcc: error: unrecognized command line option '-mssse3'
>>
>> Signed-off-by: Ilya Maximets
>
> Applied, thanks!
>
Since f35f52833c5b ("NEWS: DPDK 2.2 is now required.")
Add support for Jumbo Frames to DPDK-enabled port types,
using single-segment-mbufs.
Using this approach, the amount of memory allocated for each mbuf
to store frame data is increased to a value greater than 1518B
(typical Ethernet maximum frame length). The increased space
available in the mbuf m
??
We are trying to propose a new computation model for the ovn-northd
application. Currently, ovn-northd computes all logical data path flows for
every change. This full computation model is simple and helped verifying the
concept of logical data path design, but it also suffers performance dr
c/øæº
a¤µÃó¢jËãb;ææ~ĽàÐ6ö¤µû¿þ;?zÖ>EèXðóÀ{Gk<×[GD5/»2ÄÔ*ÊuÞÙÐ8aµÜmT½LÒô(;(zØÀqñðÜF
×ÜY%.ûñûOW`2ãrd[Jm¾ W C
¬áºiÑ
ߪʵû´óöNöw³ùÂdSsÉÖÜÏß&c
ÈùÅeùíþ<¤
¿2ìi;ð4'J-,ÊHCóÔ¼JÔÀm3ßr(а¿G´¤
¸»&
4}¯ªÒ÷$>dcÁ,NS°Ú{0õàÉÇ[3Õ\JÔ¼dA`ÊAJò#d~39þe¢`ãÂc32°ãýbÄg
Hi Ben,
Do you have more questions on this topic?
Thank you,
Babu
On Thursday 14 January 2016 06:46 PM, Miguel Angel Ajo wrote:
Hi!,
Russell Bryant wrote:
On 01/13/2016 11:30 AM, Russell Bryant wrote:
On 01/11/2016 08:19 PM, Ben Pfaff wrote:
On Tue, Jan 05, 2016 at 07:33:16PM +0530, bscha.
59 matches
Mail list logo
