[ovs-discuss] pmd segfault

[Pradeep K.S]

Hi All,

Current software version:
=
openvswitch-2.9.0-19.el7fdp.1.x86_64
dpdk-17.11-7.el7.x86_64
3.10.0-957.21.3.el7.x86_64


VM Configuration:
=
I have vm having regular 2 virtio ports, 1 vhost-dpdk , 1 sriov ports on
it. At times PMD crashes with following backtrace


(gdb) bt
#0  rte_vhost_dequeue_burst (vid=, queue_id=,
mbuf_pool=0x7f42bf7b6940, pkts=pkts@entry=0x7f45467f3770,
count=count@entry=32) at
/usr/src/debug/openvswitch-2.9.0/dpdk-17.11/lib/librte_vhost/virtio_net.c:1567
#1  0x564cce70a2e4 in netdev_dpdk_vhost_rxq_recv (rxq=,
batch=0x7f45467f3760) at lib/netdev-dpdk.c:1849
#2  0x564cce656671 in netdev_rxq_recv (rx=,
batch=batch@entry=0x7f45467f3760) at lib/netdev.c:701
#3  0x564cce62fc1f in dp_netdev_process_rxq_port
(pmd=pmd@entry=0x7f46b0358010,
rxq=0x564cd007ba90, port_no=3) at lib/dpif-netdev.c:3279
#4  0x564cce63002a in pmd_thread_main (f_=) at
lib/dpif-netdev.c:4145
#5  0x564cce6accb6 in ovsthread_wrapper (aux_=) at
lib/ovs-thread.c:348
#6  0x7f46c6804dd5 in start_thread () from /lib64/libpthread.so.0
#7  0x7f46c5c01ead in clone () from /lib64/libc.so.6
(gdb) f 3
#3  0x564cce62fc1f in dp_netdev_process_rxq_port
(pmd=pmd@entry=0x7f46b0358010,
rxq=0x564cd007ba90, port_no=3) at lib/dpif-netdev.c:3279
3279error = netdev_rxq_recv(rxq->rx, &batch);
gdb) p *(rxq->rx)
Cannot access memory at address 0x7f42bfa59b80

(gdb) f 4
(gdb) p /x *poll_list[0]->rxq
$5 = {port = 0x564cd0240670, rx = 0x7f42bfa5b200, core_id = 0x7fff,
intrvl_idx = 0x1e, pmd = 0x7f46b0358010, cycles = {0x1b870,
0x746dc}, cycles_intrvl = {0x48c76, 0x11b16, 0xe7e8, 0x180de, 0x10388,
0x124e8}}


pmd_load_queues_and_ports(pmd, &poll_list);

Let me know if you need more info, i have core file too.

Regards,
Pradeep.K.S
___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

Re: [ovs-discuss] How to filter tagged frames in bridge?

[Matthias May via discuss]

On 08/08/2019 13:43, Felipe Arturo Polanco wrote:
> The hypervisor is the one that adds the ports to the switch I specify.
> 
> Is there a way to limit vlan tags being delivered to a fake bridge perhaps? I 
> only want untagged traffic in the fake
> bridge. 
> 
> 
> On Wed, Aug 7, 2019, 2:52 AM Matthias May via discuss 
> mailto:ovs-discuss@openvswitch.org>>
> wrote:
> 
> On 06/08/2019 17:12, Felipe Arturo Polanco wrote:
> > Hello,
> >
> > This is for a hosting environment where we are using OVS bridges with 
> KVM.
> >
> > I have two interfaces bonded together with LACP and allowing two vlans.
> > VLAN 500 public and vlan 400 private.
> > The native vlan for this trunk port is Vlan 500*
> >
> > I need to find a way to limit trunk access on the VMs when they are
> > connected to my bridge.
> > If I add a tap0 interface to ovsbr0, I can see tagged traffic which is 
> not good.
> >
> > I was thinking about adding a second bridge and connect both of them
> > using a patch port but I still need to find a way to filter tagged
> > frames and only allow untagged traffic on the second bridge.
> >
> > Any ideas how can this be done?
> >
> > Thanks,
> > ___
> > discuss mailing list
> > disc...@openvswitch.org 
> > https://mail.openvswitch.org/mailman/listinfo/ovs-discuss
> >
> 
> When you add the port, set
> vlan_mode=access
> tag=500
> 
> BR
> Matthias
> ___
> discuss mailing list
> disc...@openvswitch.org 
> https://mail.openvswitch.org/mailman/listinfo/ovs-discuss
> 

I highly suggest you read the documentation regarding vlan_mode, tag and trunk.

My answer is still to set the vlan_mode to access and set the tag.
It doesn't matter if the hypervisor adds the port or someone else.
You can set a config for a port even if it is not yet part of a bridge.

BR
Matthias
___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

Re: [ovs-discuss] How to filter tagged frames in bridge?

[Felipe Arturo Polanco]

The hypervisor is the one that adds the ports to the switch I specify.

Is there a way to limit vlan tags being delivered to a fake bridge perhaps?
I only want untagged traffic in the fake bridge.


On Wed, Aug 7, 2019, 2:52 AM Matthias May via discuss <
ovs-discuss@openvswitch.org> wrote:

> On 06/08/2019 17:12, Felipe Arturo Polanco wrote:
> > Hello,
> >
> > This is for a hosting environment where we are using OVS bridges with
> KVM.
> >
> > I have two interfaces bonded together with LACP and allowing two vlans.
> > VLAN 500 public and vlan 400 private.
> > The native vlan for this trunk port is Vlan 500*
> >
> > I need to find a way to limit trunk access on the VMs when they are
> > connected to my bridge.
> > If I add a tap0 interface to ovsbr0, I can see tagged traffic which is
> not good.
> >
> > I was thinking about adding a second bridge and connect both of them
> > using a patch port but I still need to find a way to filter tagged
> > frames and only allow untagged traffic on the second bridge.
> >
> > Any ideas how can this be done?
> >
> > Thanks,
> > ___
> > discuss mailing list
> > disc...@openvswitch.org
> > https://mail.openvswitch.org/mailman/listinfo/ovs-discuss
> >
>
> When you add the port, set
> vlan_mode=access
> tag=500
>
> BR
> Matthias
> ___
> discuss mailing list
> disc...@openvswitch.org
> https://mail.openvswitch.org/mailman/listinfo/ovs-discuss
>
___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

Re: [ovs-discuss] [OVN] HA about OVN DB servers

[Numan Siddique]

On Thu, Aug 8, 2019 at 3:44 PM taoyunupt  wrote:

> Hi,Numan,
>   Thanks for you reply. Do you have advice for choosing which proxy
> software for OVN DB servers?
>
>  proxy
> software
>
> |
>
> |
>
>  --
>|
>  | |
>   OVN DB  OVN DB
>  OVN DB
>


If you want to deploy with Active/Passive, I think you need to use
pacemaker with an IPAddr2 resource as VIP.

If you want to deploy with Active/Active, then I think you can use haproxy
to load balance with a front end VIP.
Or you can configure ovn-northd and ovn-controllers to connect to the whole
cluster with multiple remotes   - tcp:IP1:6641,tcp:IP2:6612,tcp:IP3:6641.
The OVSDB IDL supports this.

Thanks
Numan


>
> Thanks,
> Yun
>
>
>
>
>
> 在 2019-08-07 22:46:39，"Numan Siddique"  写道：
>
>
>
> On Wed, Aug 7, 2019 at 3:04 PM taoyunupt  wrote:
>
>> Hi,Numan,Lucas,
>>  Does OVN support that one Northbound DB associated server Southbound
>> DBs, to overcome performance bottleneck of Southbound DB.The frame just as
>> the following pic.
>>   What is the plan of multi-master cluster of OVN DB servers?
>>
>>
>>  CMS
>>
>> |
>>
>> |
>>
>> Northbound DB
>>
>> |
>>
>> |
>>
>>  --
>>|
>>  | |
>>   Southbound DBSouthbound DB
>>   Southbound DB
>>|
>>  | |
>>chassis nodes  chassis nodes
>>chassis nodes
>>
>> Thanks,
>> Yun
>>
>
>
> Hi Yun,
>
> OVN at this point doesn't support the model you described above.
>
> ovsdb-server support RAFT with which you can have active/active deployment
> of OVN DB servers.
> Is that what you are looking at in a multi-master cluster ? Please take a
> look at "man ovn-ctl" for more details on
> how active/active cluster can be setup.
>
> With the commit [1], OVN supports transport zones. I am not sure if you
> are looking for this feature though.
>
> [1] -
> https://github.com/openvswitch/ovs/commit/3bb479d9aa52f0a25e941bb3156b5fc4ddca3c8c
>
>
> Thanks
> Numan
>
>
___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

Re: [ovs-discuss] [OVN] HA about OVN DB servers

[taoyunupt]

Hi,Numan,
  Thanks for you reply. Do you have advice for choosing which proxy 
software for OVN DB servers?


 proxy 
software
  |
  |
   
-- 
   |   
| |
  OVN DB  OVN DB
   OVN DB




Thanks,
Yun






在 2019-08-07 22:46:39，"Numan Siddique"  写道：





On Wed, Aug 7, 2019 at 3:04 PM taoyunupt  wrote:

Hi,Numan,Lucas,
 Does OVN support that one Northbound DB associated server Southbound DBs, 
to overcome performance bottleneck of Southbound DB.The frame just as the 
following pic.
  What is the plan of multi-master cluster of OVN DB servers? 


   CMS 
  |
  |
Northbound 
DB
  |
  |
   
-- 
   |   
| |
  Southbound DBSouthbound DB
Southbound DB
   |   
| |
   chassis nodes  chassis nodes 
chassis nodes  
 
Thanks,
Yun




Hi Yun,


OVN at this point doesn't support the model you described above.


ovsdb-server support RAFT with which you can have active/active deployment of 
OVN DB servers.
Is that what you are looking at in a multi-master cluster ? Please take a look 
at "man ovn-ctl" for more details on
how active/active cluster can be setup.


With the commit [1], OVN supports transport zones. I am not sure if you are 
looking for this feature though.


[1] - 
https://github.com/openvswitch/ovs/commit/3bb479d9aa52f0a25e941bb3156b5fc4ddca3c8c
 


Thanks
Numan

___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

[ovs-discuss] VLAN ID Policing

[V Sai Surya Laxman Rao Bellala]

Hello community members,

I want to send traffic to the interface with two different VLAN ID. I want
to perform ingress policing based on VLAN ID.


Thanks in advance

Regards
Laxman
___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

[ovs-discuss] [HELP] Question about handler threads in userspace datapath

[txfh2007 via discuss]

Hi all:
  I have found when using OVS-DPDK, there is still one handler thread and 
one revalidator thread. So is the handler thread necessary? As far as i know, 
in userspace , the upcall and flow put/del is done by pmd threads ? so what is 
the handler thread's job in userspace datapath ?

Thanks
Timo  



___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

[ovs-discuss] OvS segfault when starting test-dpdkr

[avatli]

Hello community

I'm trying to run the test-dpdkr application but it causes  
segmentation fault in ovs-vswitchd as below.


Aug  2 21:48:32 ulak ovs-vsctl: ovs|1|vsctl|INFO|Called as  
ovs-vsctl add-port br0 dpdkr0 -- set Interface dpdkr0 type=dpdkr

Aug  2 21:49:50 ulak test-dpdkr[31039]: EAL: Probing VFIO support...
Aug  2 21:49:50 ulak test-dpdkr[31039]: APP: Finished Process Init.
Aug  2 21:49:50 ulak kernel: [ 2172.400789] show_signal_msg: 9  
callbacks suppressed
Aug  2 21:49:50 ulak kernel: [ 2172.400795] pmd32[31036]: segfault at  
6e0088 ip 00a707c9 sp 7fb40f7fc050 error 6 in  
ovs-vswitchd[40+97d000]


ulak@ulak:~$ addr2line -e /usr/local/sbin/ovs-vswitchd 00a707c9
/home/ulak/ovs-master/lib/dp-packet.h:718

First, I tried OvS 2.11.0 and DPDK 18.11.2 version written on OVS  
releases, but the same seqfault was there, then I used the latest OVS  
and DPDK (it has a build error for RTE_COLOR_GREEN, I fixed it), and  
still the same problem


root@ulak:/home/ulak# ovs-vswitchd --version
ovs-vswitchd (Open vSwitch) 2.12.90
DPDK 19.05.0

I just followed the steps in [1], and this problem arises every time,  
my setup is as follows:


root@ulak:/home/ulak# ovs-vsctl show
5232862b-c8a0-408b-924d-41827fc7915b
Bridge br0
datapath_type: netdev
Port dpdkr0
Interface dpdkr0
type: dpdkr
Port br0
Interface br0
type: internal
ovs_version: "2.12.90"


root@ulak:/home/ulak/ovs-master/tests# ./test-dpdkr -c 1 -n 4  
--proc-type=secondary -- -n 0

EAL: Detected 4 lcore(s)
EAL: Detected 1 NUMA nodes
EAL: Multi-process socket /var/run/dpdk/rte/mp_socket_31039_4ad8f644b92
EAL: Probing VFIO support...
APP: Finished Process Init.

Client process 0 handling packets
[Press Ctrl-C to quit ...]

ulak@ulak:~$ uname -a
Linux ulak 4.4.0-157-generic #185-Ubuntu SMP Tue Jul 23 09:17:01 UTC  
2019 x86_64 x86_64 x86_64 GNU/Linux


ulak@ulak:~$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description:Ubuntu 16.04.6 LTS
Release:16.04
Codename:   xenial


Last, I was using the test-dpdkr without any problem with the OvS version 2.7.

Thanks in advance..

- Volkan

[1] http://docs.openvswitch.org/en/latest/topics/dpdk/ring/


___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss