Hello, I met a problem when I set the trust-mode for the ConnectX3-Pro 40GbE NIC.

The system information follows:

LSB Version: :core-4.1-amd64:core-4.1-noarch

Distributor ID: CentOS

Description: CentOS Linux release 7.3.1611 (Core)

Release: 7.3.1611

Codename: Core

The ConnectX3-Pro NIC information follows:

hca_id: mlx4_1

transport: InfiniBand (0)

fw_ver: 2.40.7000

node_guid: f452:1403:0095:2280

sys_image_guid: f452:1403:0095:2280

vendor_id: 0x02c9

vendor_part_id: 4103

hw_ver: 0x0

board_id: MT_1090111023

phys_port_cnt: 2

Device ports:

port: 1

state: PORT_ACTIVE (4)

max_mtu: 4096 (5)

active_mtu: 1024 (3)

sm_lid: 0

port_lid: 0

port_lmc: 0x00

link_layer: Ethernet

port: 2

state: PORT_DOWN (1)

max_mtu: 4096 (5)

active_mtu: 1024 (3)

sm_lid: 0

port_lid: 0

port_lmc: 0x00

link_layer: Ethernet

It's the first time that I have met this problem. So I don't know what to do.

What does this tip mean? Is the system version the main cause?

Waiting for your help.

Thanks.

Server:
> ib_send_bw -a -F --report_gbits
Client:
> ib_send_bw -a -F --report_gbits <serverIP>

Please let me know your results and thank you...

~Steve

Hi Steve,

Here you are.  Thanks for taking an interest.  Still getting ~45 Gb/sec on both client and server.  Here is the client output:

[root@vx01 ~]#  ib_send_bw -a -F --report_gbits vx02

---------------------------------------------------------------------------------------

                    Send BW Test

Dual-port       : OFF Device         : mlx5_0

Number of qps   : 1 Transport type : IB

Connection type : RC Using SRQ      : OFF

TX depth        : 128

CQ Moderation   : 100

Mtu             : 4096[B]

Link type       : IB

Max inline data : 0[B]

rdma_cm QPs : OFF

Data ex. method : Ethernet

---------------------------------------------------------------------------------------

local address: LID 0x3e4 QPN 0x005e PSN 0x239861

remote address: LID 0x3e6 QPN 0x004c PSN 0xed4513

---------------------------------------------------------------------------------------

#bytes     #iterations    BW peak[Gb/sec]    BW average[Gb/sec]   MsgRate[Mpps]

2          1000           0.098220            0.091887            5.742968

4          1000             0.20               0.19       6.037776

8          1000             0.40               0.39       6.071169

16         1000             0.78               0.67       5.220818

32         1000             1.53               1.43       5.576730

64         1000             3.16               3.10       6.053410

128        1000             6.20               6.16       6.012284

256        1000             12.35              12.28     5.997002

512        1000             22.67              22.47     5.486812

1024       1000             38.02              36.69     4.478158

2048       1000             42.26              42.04     2.565771

4096       1000             43.82              43.68     1.332978

8192       1000             44.63              44.63     0.681005

16384      1000             44.79              44.79     0.341728

32768      1000             45.21              45.21     0.172449

65536      1000             45.35              45.35     0.086506

131072     1000             45.45              45.45     0.043342

262144     1000             45.45              45.45     0.021670

524288     1000             45.47              45.47     0.010840

1048576    1000             45.47              45.47     0.005421

2097152    1000             45.48              45.48     0.002711

4194304    1000             45.48              45.48     0.001355

8388608    1000             45.48              45.48     0.000678

---------------------------------------------------------------------------------------

Here is the server output:

[root@vx02 ~]#  ib_send_bw -a -F --report_gbits

************************************

* Waiting for client to connect... *

************************************

---------------------------------------------------------------------------------------

                    Send BW Test

Dual-port       : OFF Device         : mlx5_0

Number of qps   : 1 Transport type : IB

Connection type : RC Using SRQ      : OFF

RX depth        : 512

CQ Moderation   : 100

Mtu             : 4096[B]

Link type       : IB

Max inline data : 0[B]

rdma_cm QPs : OFF

Data ex. method : Ethernet

---------------------------------------------------------------------------------------

local address: LID 0x3e6 QPN 0x004c PSN 0xed4513

remote address: LID 0x3e4 QPN 0x005e PSN 0x239861

---------------------------------------------------------------------------------------

#bytes     #iterations    BW peak[Gb/sec]    BW average[Gb/sec]   MsgRate[Mpps]

2          1000           0.000000            0.099141            6.196311

4          1000             0.00               0.20       6.229974

8          1000             0.00               0.40       6.265230

16         1000             0.00               0.69       5.362016

32         1000             0.00               1.47       5.727960

64         1000             0.00               3.22       6.283794

128        1000             0.00               6.34       6.191118

256        1000             0.00               12.64     6.169975

512        1000             0.00               23.08     5.634221

1024       1000             0.00               37.53     4.581582

2048       1000             0.00               42.63     2.602155

4096       1000             0.00               44.07     1.344970

8192       1000             0.00               45.04     0.687191

16384      1000             0.00               45.04     0.343602

32768      1000             0.00               45.35     0.172994

65536      1000             0.00               45.45     0.086690

131072     1000             0.00               45.52     0.043409

262144     1000             0.00               45.51     0.021699

524288     1000             0.00               45.52     0.010852

1048576    1000             0.00               45.52     0.005427

2097152    1000             0.00               45.53     0.002714

4194304    1000             0.00               45.53     0.001357

8388608    1000             0.00               45.53     0.000678

---------------------------------------------------------------------------------------

Please let me know if you want any other info and I will send it straight away.

Regards,

Eric

Hi,

Is ASAP2 OVS Offload support OpenStack Live Migration? If not, which ASAP2 mode should I use, OVS Acceleration or Application Acceleration (DPDK Offload)? How to have H/W LAG (with LACP) on each of those three modes?

Best regards,

Hello.

I have Microsoft Windows 2012 R2 cluster and some nodes have ConnectX-3 adapters and some nodes have ConnectX-4 Lx adapters.

There is RoCE connectivity between nodes with ConnectX-4 Lx adapters, but there isn’t connectivity between nodes with different adapters.

I think it’s because ConnectX-3 adapters use RoCE  1.0 mode, but ConnectX-4 Lx adapters use RoCE 2.0 mode.

I tred to change RoCE mode from 1.0 to 2.0 for ConnectX-3 adapters by “Set-MlnxDriverCoreSetting -RoceMode 2”, but had warning – “SingleFunc_2_0_0: RoCE v2.0 mode was requested, but it is not supported. The NIC starts in RoCE v1.5 mode” and RoCE connectivity doesn’t work.

What the best way to fix my problem ?

ConnectX-3 adapters don’t work in RoCE 2.0 mode at all ?    How about new FW ? Today I have WinOF-5_35 with FW:

Image type:      FS2

FW Version:      2.40.5032

FW Release Date: 16.1.2017

Product Version: 02.40.50.32

Rom Info:        type=PXE version=3.4.747 devid=4099

Device ID:       4099

Description:     Node             Port1            Port2            Sys image

GUIDs:           ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff

MACs: e41d2ddfa540     e41d2ddfa541

VSD:

PSID:            MT_1080120023

I don’t find way to change to RoCE 1.0 mode for ConnectX-4 Lx adapters for Microsoft Windows environment.

Hi all,

After installing MLNX_OFED_LINUX-4.4-1 on Ubuntu 18.04 (kernel 4.15.0-24) as "$ mlnxofedinstall --force --without-dkms --with-nvmf" I'm trying to use RDMA tools, but

- modprobe on nvme_rdma fails with "nvme_rdma: Unknown symbol nvme_delete_wq (err 0)"

- modprobe on nvmet_rdma fails with "nvmet: Unknown symbol nvme_find_pdev_from_bdev (err 0)"

What am I doing wrong, please?

I see two kernel modules loaded: nvme and nvme_core.

This is Mellanox MCX516A-CCAT ConnectX-5 EN Network Interface Card 100GbE Dual-Port QSFP28.

Any inputs will be greatly appreciated

Thank you

Dmitri Fedorov

Ciena Canada

Hi,

I have a ConnectX-3 card with two physical ports, each port has 4 RX queues. I use another machine to generate sample traffic and send to both ports. However, there is only one queue that are receiving packets (from one IRQ channel), for BOTH physical ports.

How can I distribute the traffic to all queues (different IRQ channel)? Or at least, two physical ports should have different queues.

Here is the output from cat /proc/interrupts  | grep mlx4 :

 29:       2787       2570       2500       3093  IR-PCI-MSI 524288-edge      mlx4-async@pci:0000:01:00.0
 30:          0          0         66         11  IR-PCI-MSI 524289-edge      mlx4-1@0000:01:00.0
 31:          9          0         88          0  IR-PCI-MSI 524290-edge      mlx4-2@0000:01:00.0
 32:        124         13          0          0  IR-PCI-MSI 524291-edge      mlx4-3@0000:01:00.0
 33:     385333          0    2517273          0  IR-PCI-MSI 524292-edge      mlx4-4@0000:01:00.0
 34:          0          0          0          0  IR-PCI-MSI 524293-edge      mlx4-5@0000:01:00.0
 35:          0          0          0          0  IR-PCI-MSI 524294-edge      mlx4-6@0000:01:00.0
 36:          0          0          0          0  IR-PCI-MSI 524295-edge      mlx4-7@0000:01:00.0
 37:          0          0          0          0  IR-PCI-MSI 524296-edge      mlx4-8@0000:01:00.0

The packet counts from ifconfig show that both ports receive packets.

I use Ubuntu 18.04 64 bit with kernel v4.15. I use upstream driver from the kernel (because I need to run my XDP program on the NIC. However, in the test above, XDP is not running), and havent install Mellanox OFED yet. i I believe the card should work fine with only the drivers from the kernel(?).

> Service for this Hardware and FW has ended. So it will not be hosted on our site.

Thank you for your help. Seems this new firmware is no newer than the one I already have, unfortunately.

Those are 10+ years old cards... no surprise they are difficult to put back in service.

Cordially & thanks again for the great support,

I have a limited budget, and I want to buy a high speed network interface cards and a switch to connect multiple PCs into a local network (just a couple of meters away from one another) for HPC research.

Some of the key points for me are:

• It should be compatible with Windows
• Speed preferably 56 gigabit
• Needs to be able to sustain a 100% load at all times
• Non-managed Switch
• No SFP(+) uplinks

Taking all of that into account what models should I look to buy? Perhaps some older models from a couple of years ago?

What are some caveats in building a system like that that I should consider?

What should I expect in terms of latency?

Hi All,

I don't know if someone already posted something about this.

I just want to share that when we upgraded our SN2100 to v3.6.8004, the GUI doesn't load. Luckily it can still be accessible via SSH and I changed the next boot partition and it rebooted fine. I downgraded the partition from a working version and now we can manage the switch both via GUI and CLI.

Hope this helps someone who wants to upgrade to this version.

I have encountered this question, too.

It was because of the ucx do not compile with cuda.(The mlnx install the default ucx).

When I recompile the ucx with cuda and reinstall it ,It works.

I've made a feeble attempt to utilise Header Data Split (HDS) offload on Connect-X 5 adapters, by creating the striding WQ context with a non-zero log2_hds_buf_size value. However, the hardware won't have it and reports back bad_param error with syndrome 0x6aaebb.

According to an online error syndrome list, this translates to human readable as:

create_rq/rmp: log2_hds_buf_size not supported

Since the Public PRM does not describe HDS offload, I'm curious to whether certain preconditions need to met for this offload to work, or if this is a known restriction in current firmware? I'd also like to know if it's possible to configure the HDS "level", that is where the split happens (between L3/L4, L4/L5, ...).

The way I'd envision this feature to work is to zero-pad the end of headers up to log2_hds_buf_size, placing the upper layer payload at a fixed offset for any variable-size header length.

Hello Eric -

   I hope all is well...

You won't achieve a line rate of 56G/s because the NIC is: MCB193A-FCAT MT_1220110019 and your PCEi is 2.0

And the release notes for you FW state:

Connect-IB® Host Channel Adapter, single-port QSFP, FDR 56Gb/s,PCIe3.0 x16, tall bracket, RoHS R6

So getting ~45 ~48 Gb/s is good.

Have a great day!

Steve

is it possible to configure RoCE v2 with Connectx-4 card without MLNX_OFED? can someone please share info if there is any guide/doc available to configure with Linux drivers and packages?

I tried to do with drivers and packages but I am not able to succeed. When I used MLNX_OFED, RoCE is configured successfully.

Hello Mellanox,

I have ConnectX-3 QSFP Dual Port CX354A  Adapter with Windows 7 x64 Pro on PC, I would like to make throughput test between QSFP ports. For this connected 1m fiber cable between ports and set CX354A in Ethernet mode in Device Manager. I also manualy set different IP addresses for 2 Ethernet adapters with the same IP mask:

IP1: 192.168.0.1

mask: 255.255.255.0

IP2: 192.168.0.2

mask: 255.255.255.0

I tried make tcp iperf3 test, but I had only 12 Gbit/s instead of 40 Gbit/s. In your documentation where ib_send_bw  utility was recommended for testing the performance.

How make test with ib_send_bw between two Ethernet interfaces on one PC?

For example: to set interface in iperf3 I can choose with key -B: iperf3 -B 192.168.0.1.

Thank you.

--

With Best Wishes,

Dmitrii

Hi everybody !

how are connectx-5 GID created/initialized ? they disappeared after ofed upgrade...

ibv_devinfo -v neither show_gids display any GID...

Any ideas

regards, raph

after upgrade to MLNX_OFED_LINUX-4.4-1.0.0.0-debian9.1-x86_64

scisoft13:~ % sudo show_gids

DEV     PORT INDEX   GID                        IPv4 VER DEV

---     ---- -----   ---                        ------------ --- ---

n_gids_found=0

before upgrade

scisoft13:~ % sudo show_gids

DEV     PORT INDEX   GID                        IPv4 VER DEV

---     ---- -----   ---                        ------------ --- ---

mlx4_0  1 0       fe80:0000:0000:0000:526b:4bff:fe4f:be21                 v1 enp131s0

mlx4_0  2 0       fe80:0000:0000:0000:526b:4bff:fe4f:be22                 v1 enp131s0d1

mlx5_0  1 0       fe80:0000:0000:0000:526b:4bff:fed3:d164                 v1 enp130s0f0

mlx5_0  1 1       fe80:0000:0000:0000:526b:4bff:fed3:d164                 v2 enp130s0f0

mlx5_0  1 2       0000:0000:0000:0000:0000:ffff:c0a8:030d 192.168.3.13    v1 enp130s0f0

mlx5_0  1 3       0000:0000:0000:0000:0000:ffff:c0a8:030d 192.168.3.13    v2 enp130s0f0

mlx5_1  1 0       fe80:0000:0000:0000:526b:4bff:fed3:d165                 v1 enp130s0f1

mlx5_1  1 1       fe80:0000:0000:0000:526b:4bff:fed3:d165                 v2 enp130s0f1

n_gids_found=8

thank you for helping ! it is strange because everything was ok before

upgrade.

- ping is ok

- /sys/class/infiniband/ etc exists and is populated but gids...

after upgrade to MLNX_OFED_LINUX-4.4-1.0.0.0-debian9.1-x86_64

scisoft13:~ % sudo show_gids

DEV     PORT INDEX   GID                        IPv4 VER DEV

---     -

scisoft13:~ % cat /sys/class/infiniband/mlx5_0/ports/1/gids/0

0000:0000:0000:0000:0000:0000:0000:0000

scisoft13:~ % cat /sys/class/infiniband/mlx5_0/ports/1/gids/1

0000:0000:0000:0000:0000:0000:0000:0000

scisoft13:~ % cat /sys/class/infiniband/mlx5_0/ports/1/gid_attrs/types/0

cat: /sys/class/infiniband/mlx5_0/ports/1/gid_attrs/types/0: Invalid argument

same issue on 2 servers with connectx-5 EN 100Gb/s optical link and connectx-3 40GBb/s  copper link

ofed install without issue

Hi Raphael,

Thank you for the information, it looks as an unexpected behaviour related to the driver and this specific Operating system.

For us to continue and investigate it please send an email to support@mellanox.com and open a support ticket with all the details.

Thank you,

Karen.

Thanks a lot for the reply. It solved the above issue but after running mpirun, i do not see any latency difference with and without GDR

My Questions :

1. Why I do not see any latency difference with and without GDR. ?
2. Does below sequence or steps correct ? Does it matter for my Question 1

Note: I am having single GPU on both host and peer. Iommu is disabled.

## nvidia-smi topo -m

           GPU0    mlx5_0  mlx5_1  CPU Affinity

GPU0     X      PHB     PHB     18-35

mlx5_0  PHB      X      PIX

mlx5_1  PHB     PIX      X

Steps followed are:

1. Install CUDA 9.2 and add the library and bin path in .bashrc

2. Install latest MLX OFED

3. Compile and Install nv_peer_mem driver

4. Get UCX from git. Configure UCX with cuda and  Install UCX

5. Configure Openmpi-3.1.1 and install it.

./configure --prefix=/usr/local --with-wrapper-ldflags=-Wl,-rpath,/lib --enable-orterun-prefix-by-default --disable-io-romio --enable-picky --with-cuda=/usr/local/cuda-9.2

6. Configure OSU Benchmarks-5.4.2 with cuda and install it

./configure prefix=/root/osu_benchmarks CC=mpicc --enable-cuda --with-cuda=/usr/local/cuda-9.2

Run mpirun. I do not see any latency difference with and without GDR.

Thanks for your Help.