Hi David,
This will likely be the same issue reported in this discussion:
http://openvswitch.org/pipermail/discuss/2016-March/020551.html
The issue is fixed in DPDK, in the commit 9a0615af7746("virtio: fix restart")
2016-04-06 13:17 GMT-07:00 David Evans :
> Hi openvswitch,
>
> I am trying to
2016-04-26 9:08 GMT-07:00 Traynor, Kevin :
> > -Original Message-
> > From: discuss [mailto:discuss-boun...@openvswitch.org] On Behalf Of
> > Kochba, Alon
> > Sent: Tuesday, April 26, 2016 4:38 PM
> > To: Ben Pfaff ; Yi Ba
> > Cc: b...@openvswitch.org
> > Subject: Re: [ovs-discuss] ovs ge
Short term solution:
Yes, returning "ODP_FIT_TOO_LITTLE" from check_expectations() isn't enough to
prevent a flow from being installed in the kernel datapath. OVS userspace,
since the introduction of megaflows, handles compatibility with older kernels
in a different way: the kernel will refuse
Hi Mattias,
thanks for reporting this!
The delayed flushing logic was implemented for performance reasons: the
idea is to reduce the calls to rte_eth_tx_burst(). The tx queue is supposed
to be flushed also in netdev_dpdk_rxq_recv(), but that doesn't always work
as intended, because we rely on the
Hi Subramani,
I appreciate the feedback on the userspace connection tracker, thanks
Yes, the ICMP state is identified by the tuple (SRC IP, DST IP, ID). This
mimics the behaviour of FreeBSD's pf.
I guess I'd be possible to have more intelligence on ICMP connections by
tracking sequence numbers
ck. To avoid it used below condition */
>+if ((conn->up.key.nw_proto == IPPROTO_ICMP)
>+&& reply
>+&& (icmp->icmp_type == ICMP_ECHO_REQUEST)) {
>+
>+return CT_UPDATE_INVALID;
>+}
>
> if (reply && conn->
Hi, thanks for the patch!
Generally we send patches to d...@openvswitch.org, it's easier for reviewers
to read them,
would you mind sending next version there?
Couple of comments inline, otherwise this looks good to me
2016-05-16 22:28 GMT-07:00 AZ :
> When we're adding the port into ovs bridge
d for inclusion:
https://github.com/ddiproietto/ovs/tree/userconntrack_20160516
Thanks,
Daniele
>
>Regards,
>Soumyadeep
>
>-----Original Message-
>From: Daniele Di Proietto [mailto:diproiet...@vmware.com]
>Sent: Tuesday, May 17, 2016 6:31 AM
>To: Soumyadeep Chowdhury (Cisco)
On 20/05/2016 10:35, "Daniele Di Proietto" wrote:
>Hi Soumyadeep,
>
>
>
>On 18/05/2016 22:58, "soumyadeep.chowdh...@wipro.com"
> wrote:
>
>>Hi Daniele,
>>
>>Its much cleaner approach.
>>Thanks for the update.
>
>Tha
Hi Sugesh,
Thanks for working on this, I think this is going to be very useful.
You may want to take a look at ovs-bugtool (in utilities/bugtool), it
performs many of the tasks you're mentioning. It should be easily
extendable with plugins, I can imagine a few DPDK (or OVS-DPDK) specific.
What d
2016-05-18 0:24 GMT-07:00 Bhanu Chander Gaddoju :
> Hi Mark,
>
>
>
> Thank you for your response.
>
> We are seeing single PMD thread even after setting pmd-cpu-mask
> to 3 as below after starting the vswitchd daemon.
>
>
>
> * ovs-vsctl set Open_vSwitch . other_config:pmd
Hi Mechthild,
Have you tried with the latest master?
I suspect this might have already been fixed by b59cc14e032d("netdev-dpdk:
Use instant sending instead of queueing of packets."). In this case I need
to backport that on branch-2.5 as well.
Thanks,
Daniele
2016-07-12 17:23 GMT-07:00 Mechthi
r dpdk ports is still 21. I saw another patch
> which introduces n_txq which I haven’t tried yet but will do soon….
>
That n_txq is only used for testing, the number of transmission queues is
determined by the datapath or by qemu in case of vhost.
>
>
> Thank you very much
I can reproduce this too
With -march=native, if the CPU has CRC32 extensions we use a different hash
function. I suspect the dhcp options are output on the packet in a
different order because of this. Perhaps we should make the test agnostic
of the order, or order the options on the DHCP packet.
2016-09-28 2:57 GMT-07:00 Wojciechowicz, RobertX <
robertx.wojciechow...@intel.com>:
> On Fri, Sep 16, 2016 at 05:11:11PM +, John Phillips wrote:
> > When I try to set other_config:n-dpdk-rxqs to 64 with two intel niantic
> dpdk
> > ports on a single dpdk bridge, the bridge instance will 'cras
Could you share more details about the setup, please?
Which version of Open vSwitch are you using? (the exact commit would be
helpful)
What's the controller?
What's the distro?
Thanks,
Daniele
2016-10-11 1:44 GMT-07:00 Ali Volkan Atli :
> Hi OvS
>
> I'm getting the same exception each time (dp
2016-10-13 10:33 GMT-07:00 Géza Gémes :
>
>
> 2016-10-13 19:28 GMT+02:00 Géza Gémes :
>
>> Hi,
>>
>> Sorry for cross-posting, but I feel this might be an interesting topic
>> for the dev list as well.
>>
>> I've recreated my setup with qemu VM instead of lxc container and the
>> situation is the s
Hi Sharmila,
``ovs-vsctl add-br br5'' tells OVS to create a bridge that uses
a “system” dpif, i.e. a datapath created by the OVS linux kernel
module (not a userspace datapath, usable with DPDK devices).
The error in the log suggests that the OVS kernel module is not
loaded.
Daniele
> On 17 Mar
> On 17 Mar 2015, at 13:49, Radhakrishnan, Sharmila
> wrote:
>
> Hi Daniele,
>
> I went through e OVS installation in userspace. In the Userspace install it
> is given, we install everything except the kernel modules and the add the
> bridges using the below commands:
>
> ovs-vsctl add-br b
> On 4 May 2015, at 10:59, Gergely Pongracz
> wrote:
>
> Hi,
>
> We ran into some issues using the DPDK based OVS switch (2.3.90). We used
> pktgen for traffic generation.
Hi Gergely,
thanks for sharing these results
> *1. Multiple flows -> degrading performance*
>
> Everything looks nice
Hi Alexander,
> On 13 May 2015, at 17:30, Alexander Shalimov wrote:
>
> Hi all,
>
> I use the latest OVS (master branch) running only DPDK ports. I have
> connected an OpenFlow controller that iteratively sends LLDP packets via
> OpenFlow Packet Out messages across all ports.
>
> I see that
On 19/05/2015 10:24, "钢锁0310" wrote:
>
>
>I setup ovs-dpdk by following INSTALL.DPDK.md
>
>
>[root@25 tools]# ./dpdk_nic_bind.py --status
>
>Network devices using DPDK-compatible driver
>
>:00:19.0 'Ethernet Connection I217-LM' drv=igb_uio unused=
uot;37f926ae-5388-481e-a9da-4447904c1bf6"]}}},"O
>pen_vSwitch":{"d384f9bb-072c-4505-afc8-f136fb2e860d":{"new":{"manager_opti
>ons":["set",[]],"bridges":["uuid","d64a8e1a-0e69-4cbd-9081-332e5a4e04e6"],
>"statis
This might be related to offloading features.
Could you try again with this qemu "-device" option and let us know?
-device
virtio-net-pci,netdev=net1,csum=off,gso=off,guest_tso4=off,guest_tso6=off,g
uest_ecn=off
Kevin, do you this we should mention this in INSTALL.DPDk.md?
On 20/05/2015 10:39
On 20/05/2015 15:21, "Traynor, Kevin" wrote:
>
>
>> -Original Message-----
>
>> From: Daniele Di Proietto [mailto:diproiet...@vmware.com]
>
>> Sent: Wednesday, May 20, 2015 1:47 PM
>
>> To: 钢锁0310
>
>> Cc: 通天晓0280; discuss; Traynor
On 21/05/2015 15:48, "Traynor, Kevin" wrote:
>
>> -Original Message-----
>> From: Daniele Di Proietto [mailto:diproiet...@vmware.com]
>> Sent: Wednesday, May 20, 2015 4:12 PM
>> To: Traynor, Kevin
>> Cc: ???0280; discuss; ??0310
>> Subject:
04:48:42.692Z|1|dpdk(cuse_thread3)|INFO|vHost Device
>'tap0' (0) has been added
>2015-05-21T04:48:51.750Z|2|dpdk(cuse_thread3)|INFO|vHost Device
>'tap1' (1) has been added
>
>
>
>qemu create tap with default name 'tap[id]' ad
Thanks for the report!
I sent out a fix:
http://openvswitch.org/pipermail/dev/2015-May/055675.html
Would you mind trying again with the above patch?
Thanks!
On 28/05/2015 03:21, "차은호" wrote:
>
>On Thu, May 28, 2015 at 12:46 AM, Ben Pfaff
> wrote:
>
>On Wed, May 27, 2015 at 10:54:04AM +0900,
Hi,
OVS 2.3.2 supports DPDK 1.6, as stated in INSTALL.DPDK.
Many APIs have changed in DPDK in a backward incompatible
way.
Also, DPDK support in OVS 2.3.2 is experimental. It would be
better to use the code from branch-2.4
Daniele
On 19/06/2015 18:08, "Serguei Bezverkhi (sbezverk)"
wrote:
>H
Hi,
OVS 2.3.2 supports DPDK 1.6, as stated in INSTALL.DPDK.
Many APIs have changed in DPDK in a backward incompatible
way.
Also, DPDK support in OVS 2.3.2 is experimental. It would be
better to use the code from branch-2.4 (with DPDK 2.0)
Daniele
On 22/06/2015 23:30, "Sundar Ramakrishnan" wr
The commands you're typing seem correct.
Which version of OVS are you using? How are you compiling it?
On 23/06/2015 08:23, "Rapelly, Varun" wrote:
>Hi All,
>
>I¹m a newbie to openvswitch and I followed the below link to install ovs
>with DPDK on RHEL7 machine.
>
>[root@artha ~]# uname -a
>Li
in this case?
>
>Regards,
>Varun
>
>-Original Message-
>From: Rapelly, Varun
>Sent: Wednesday, June 24, 2015 10:07 AM
>To: 'Daniele Di Proietto'
>Cc: b...@openvswitch.org
>Subject: RE: [ovs-discuss] No Ethernet devices found with I350
>
>Hi Dan
ts -Qunused-arguments... no
>
>checking whether gcc -std=gnu99 accepts -Wno-unused... yes
>
>checking whether gcc -std=gnu99 accepts -Wno-unused-parameter... yes
>
>checking target hint for cgcc... x86_64
>
>checking whether make has GNU make $(if) extension... yes
>
>con
h
>and/or dpdk?
>
>
>3. Also, I am unable to use the ofctl commands. Following is what I get -
>
>
>[root@test1 ~]# ovs-ofctl show br23
>ovs-ofctl: br23 is not a bridge or a socket
>[root@test1 ~]# ovs-vsctl show
>d8cfc890-a4fc-4398-869c-59c5412a3e79
>Bridge "br23"
>
Hi,
Thanks for the very detailed report! I've sent two
patches to the mailing list that should address the
issues. Would you mind testing them?
More comments inline
On 15/07/2015 11:49, "Stokes, Ian" wrote:
>Hi All,
>
>I¹ve been investigating a segmentation fault caused by the incorrect
>se
I regularly start ovs with dpdk while traffic is flowing.
Is there anything suspicious in the ovs-vswitchd log?
Have you tried other DPDK apps (e.g. l2fwd)?
On 25/09/2015 17:28, "David Evans" wrote:
>Hi all,
>
>
>Has anyone seen that when starting ovs (with dpdk) that the port will
>show many
nect the dpdk ports to other bridges in the
> same switch.
>
>
>Could it have leaked out all it's packet buffers?
>Or the threads be busy wait somewhere else because of a race condition
>with all those PMD's?
>Maybe a thread synchronize, or quiescing issue?
>
>
>C
On 30/09/2015 04:44, "David Evans" wrote:
>Hi OVS (Ben particularly :) )
>
>How do i get OVS to assign ports to the PMD's that I choose?
>
>If i have say 6 or 12 ports and i want them distributed evenly across a
>mask of 12 or more cores on a multi node numa system / or where i know
>the NICs a
Hi,
Currently OVS behaves like most DPDK apps and examples: it uses the port
ID to
name an interface.
This is less than ideal: the PCI id might be a better alternative, but the
DPDK
community is currently considering other options to identify an interface.
I
think OVS should be consistent with o
Hi,
Currently every DPDK mbuf in OVS has the `refcnt` set to one. Output to
multiple ports is handled by making a copy of the packet's payload (see
`may_steal` in dp_netdev_execute_actions(), and in netdev_send()).
You're right, having a `refcnt` != 1 might be necessary to use
rte_ipv4_fragment_p
use
>case.
>
>Thanks for replying
>
>Dave.
>
>> On Oct 20, 2015, at 1:33 PM, Daniele Di Proietto
>> wrote:
>>
>> Hi,
>>
>> Currently every DPDK mbuf in OVS has the `refcnt` set to one. Output to
>> multiple ports is handled by making a
Hi,
Using OVS with DPDK on BSD is not supported at the moment.
The current OVS code assumes that the DPDK is compiled with vhost support,
which is not supported on BSD. Also, some of the APIs that we use in
netdev-dpdk.c are linux specific (e.g. pthread_setaffinity_np).
I suppose it won't be to
;Federico
>
>-Original Message-
>From: discuss [mailto:discuss-boun...@openvswitch.org] On Behalf Of Ben
>Pfaff
>Sent: Tuesday, January 26, 2016 7:13 AM
>To: Daniele di Proietto
>Cc: discuss@openvswitch.org
>Subject: Re: [ovs-discuss] dpdk watchdog stuck?
>
>Daniele
On 29/01/2016 15:14, "Ben Pfaff" wrote:
>On Fri, Jan 29, 2016 at 12:02:04PM +, sourabh.ban...@wipro.com wrote:
>> Hi OVS Folk,
>>
>> I checked out OVS 2.5 branch code from git hub and building ovs 2.5
>>with DPDK 2.2.0 on Centos OS 7, kernel 3.18.22 and its building
>>successfully with bel
when I tried to get it from ovs-2.5 branch its not there for
>>user space handling. Does he planning something on this separately. Any
>>idea?
>>
>>
>> I am looking for conntrack in OVS+DPDK setup.
>> So, What is the current state of this development (I am referrin
Hi,
As part of his MTU series, Mark sent a patch to do this:
http://openvswitch.org/pipermail/dev/2016-February/066406.html
On 19/02/2016 10:40, "David Evans" wrote:
>Hey Pravin & OVS wizards.
>
>Just found something fun. If you're updating to dpdk 2.2 the support is
>better for tx handling wh
Thanks for reporting this problem
I think that if a user wants to remove a port, we should just remove the
port.
Do you have any thought on this Kevin? Do you know why this check was
introduced in the first place?
Thanks
On 24/02/2016 07:33, "discuss on behalf of Jan Wickbom"
wrote:
>Hi,
>If
This appears to be a limitation in the DPDK library:
I see possible two solutions:
1) Recompile DPDK with CONFIG_RTE_MAX_ETHPORTS > 32
2) Use "--pci-blacklist" DPDK option to limit the number of ports detected
by DPDK
On 03/03/2016 10:10, "discuss on behalf of John Wei"
wrote:
>I am setting up
I'd like to be proven wrong, but I don't think that we can do much about
this from the OVS side, since it's qemu that starts the vhostuser
connection.
On 19/02/2016 02:51, "discuss on behalf of gong_ys2004"
wrote:
>
>
>
>
>Hi,
>I am configuring dpdk ovs with openstack. The OVS is 2.5.0, dpdk is
The documentation for the DPDK sub option can be found, for example, here:
http://dpdk.org/doc/guides-2.2/testpmd_app_ug/run_app.html
In this specific case, no, I don't think that you can use wildcards
On 05/03/2016 17:33, "discuss on behalf of John Wei"
wrote:
>On the pci-blacklist, does it s
The hugepages mapping is done in DPDK as part of the library
initialization, so this looks like a DPDK issue.
Would you mind reporting that to the dpdk-dev list?
On 10/03/2016 11:44, "discuss on behalf of John Wei"
wrote:
>I was trying to run multiple copies of ovs-vswitchd on different
>contai
On Jun 10, 2014, at 2:45 PM, Keqiang He wrote:
> Bug/Error :
> insmod ./datapath/linux/openvswitch.ko, I got "Error: could not insert module
> ./datapath/linux/openvswitch.ko: Unknown symbol in module”.
>
> Running "dmesg tail", I got the following -
> 6983759.520274] openvswitch: Open vSwitch
Hi Daniel,
On 9/25/14, 10:34 AM, "Daniel Badea" wrote:
>I encountered the following issue: ovs-vswitchd with crashes when adding
>a dpdk port.
>
>Building ovs with dpdk support is not straightforward because:
> - ovs doest not build with the latest dpdk version 1.7.1
We're regularly building
vs/commit/1b7a04e05b6fd15fb4caa7ec20d09ba7
aa73399f) commit. I'm glad that you can confirm it's fixed.
>
>What is the expected release date for OVS 2.4? (I'm interested in the dpdk
>vhost support)
OVS 2.4 will likely be released before the end of the year. Hopefully we
Hi,
I've been able to reproduce the issue. A fix has been pushed to master
(commit 58f195398e4d).
Can you please confirm this solves the problem for you?
Thanks for the bug report,
Daniele
On 10/1/14, 5:55 AM, "Daniel Badea" wrote:
>I¹m trying to configure a flow between a DPDK port and a VM
On 10/8/14, 2:13 PM, "MCGRATH, GIL" wrote:
>Being new to this dpdk stuff sorry for being a bit naïve. But I do need to
>some help.
>
>The documentation says vswitchd will print the number of dpdk devices
>found?
>
>What is the exactly is the command I use to do this? I already am running
>the
Hi,
On 10/8/14, 4:50 PM, "Shivapriya Hiremath" wrote:
>Hi ,
>
>
>I am not able to add the dpdk port to a bridge. I get the following error.
>
>
>
>./utilities/ovs-vsctl add-port br0 dpdk0 -- set Interface dpdk0 type=dpdk
>ovs-vsctl: Error detected while setting up 'dpdk0'. See ovs-vswitchd lo
Hi,
The warning is not particularly bad. A "revalidator" might fail to delete
a flow for good reasons (especially under heavy load).
It would be interesting to know more about the problems you're facing:
- What is your setup? (output of ovs-vsctl show)
- Are you working with OVS from current mast
58 matches
Mail list logo