Does UFM use DNS or IP to communicate with the switches?
Thanks in advance
Does UFM use DNS or IP to communicate with the switches?
Thanks in advance
There are 2 layers of communication
For inband communication and statistics UFM sends standard MADs and receives specific InfiniBand traps in return.
For management purposes, Mellanox switches have an embedded UFM agent.The UFM Server sends a multicast notification to MCast address 224.0.23.172, port 6306. The switch management replies to UFM (via port 6306) with a unicast message that contains relevant information about the switch (IP,GUID,Firmware etc...)
Afterwards the information is being exchanged via XML
Is there anyone that can help with my follow up query?
We have another server that started reporting the same issue and had the file appear on it last week, after that the server started crashing repeatedly. It appears your file is being used for "bad things" of some description so I'd like to confirm hashes and the like.
thanks
Hi, I'm Vincenzo Greco and I've installed in my company OpenStack(Juno) with 5 compute node, 1 network node, 1 controller node and 1 storage node using CentOS 7. All node have Infiniband and they work perfectly. My question is: is possible to virtualize the instance interface with infiniband?And, if it is possible, how can I make it? thank you so much for your attention and for your work!
Hi! I have a same problems and give up it now!
Because there is a compatibility issue on core driver.
Both IB and ETH require Mellanox core driver, but difference in version and compatibility. If Mellanox release common core driver for both IB and ETH then it's possible...:)
HI! Everybody!
I found ESXi MFT tools has s change port mode capability. But it can be support ConnectX-3 or above only. There is nothing to do for 2nd port ethernet function with 1.8.2.4 VPI driver. I don't know why mlnx-config function is included in ESXi MFT tools.
Is there anyone who tried mlnx-config with old 1.6.2 ethernet driver?
Hello,
I installed OFED_MLNX, Accelio and nbdx packages. My card is Connect X3 PRO.
I can see /dev/nbdx0 from client and issue READ/WRITE I/O against /dev/nbdx0.
However, I got different data when I read back the data I had written onto /dev/nbdx0.
You wouldn't notice this problem if you just run "fio" test, because "fio" does not check data integrity by default.
Anyone has seen this problem?
Thanks,
Ted
This can happen if you (accidentally) install the 32 bit version of Dell Open Manage Server Administrator on a server running 64 bit Windows Server 2008 R2. Ask me how I know:-)
Illumos is increasing in popularity, with currently available distributions such as:
SmartOS by Joyent - https://smartos.org/
OpenIndiana - OpenIndiana
NexentaStor (CE and Enterprise Edition) - Home - Nexenta
and others: Distributions - illumos - illumos wiki
My use case specifically, is SmartOS and Nexenta. I use SmartOS for virtualization, and Nexenta for NAS file storage hosting.
Our internal network is built with Mellanox Switches, specifically the SX1012's. With today's growing need and desire for high performance, low latency bandwidth, it won't be long before the 10G cards currently in use will be maxed out. It would be ideal if a Illumos driver for the Mellanox ConnectX-3 cards were available. Once a driver is made available and public, it can be certified by Illumos and other companies such as Nexenta.
This is a huge opportunity for Mellanox and their customers, as they can corner the Illumos market. Currently there are no 40G solutions for Illumos, and specifically Nexenta. Mellanox could stand to gain a lot of additional traction in the storage market, and I know I would be very happy to introduce 40G to our NAS appliances.
Thank you for your time.
Best Regards,
Kyle
I'd like that as well.
I would not get your hopes up for Nexenta support though. Even the old ConnectX-2 cards got their drivers removed from Nexenta v4 (and were never officially supported in the first place, even in v3). Nexenta did promise support for them in v4 way back in 2013, but apparently that got scratched and they never delivered.
Hello,
I'm putting in (or planning on putting in) a Ceph Cluster with my Proxmox (Virtual Hypervisor) in my home network. Between my 3 Ceph servers I want to use Infiniband and saw some decent prices on ebay. I'd like to learn more about the types of switches available and I'm hoping this community can provide some tips for me.
I'm looking at using DDR (20Gbps) Nics in my servers. Are there 8 port Infiniband switches you can suggest I look for on ebay that would work? Something that isn't too loud. I see lots of larger 24 port switches but for my small setup 8 ports would be more than enough.
I also found an Flextronics F-X430066 8-Port 4X SDR InfiniBand External Switch for a great price but I'm assuming that this switch would limit me to 10Gbps...is this a correct assumption?
Any advice you can give would be greatly appreciated.
Thank you,
I have a pair of SuperMicro servers, each with dual port ConnectX-3 EN cards. Each of these servers is connected to a SX1012, and between the SX1012s is a 40Gb trunk all using Mellanox cabling. The OS/cards report their own link is up at 40Gb. So in theory end to end there is a clean 40Gb path but results with netperf and iperf are telling me anything but. Despite disabling PF, dialing up socket buffers, tcp window sizes, minimum segment sizes, stack send queue, forcing interrupt cpu affinity, the best I can pull out of these cards with a 1500 MTU is about 3.5Gb/s. Only bumping to a 9000 MTU and setting all access and trunk interfaces to the same on the SX1012s, gets me a jump in performance. However that still only brings me a consistent max throughput of 14Gb/s. For comparison, I have four other SuperMicro servers with Intel 82599 10Gb interfaces. Those are connected to the same SX1012 switches using the 10/40 fanout cables and with not nearly the same amount of goofing around or crazy network stack settings they're each able to push a near-wire speed of 9.3Gb/s on a stock MTU of 1500.
The ConnectX-3 cards have been flashed to the 2.1.5 firmware and drivers. The SX1012 switches are running the latest 3.4.1120 MLNX-OS.
The throughput numbers are so odd they don't really give me much of a clue what is going on. They're certainly not secretly negotiating 10Gb while reporting 40. Looking at the SX1012 side I don't see any errors or discards occurring on the relevant interfaces.
We'll see I suppose. I also submitted the request to Nexenta, who have filed it internally as: NEX-3382 - Feature Request: 40G Card Support on the NexentaStor/HCL
Hi,
Configuration:
Host: RHEL7.0
Mellanox Card: ConnectX-3 Pro FDR InfiniBand + 40 GigE (MX354A-FCCT)
Mellanox Driver: MLNX_OFED_LINUX-2.4-1.0.4-rhel7.0-x86_64
Mellanox Switch: SX1012
I have one server connected to the switch with above config. After installation of OFED driver, I have set port type as "eth" ports and configured the IP for the port. However, I am not able to see the LED's lighting up on the switch as well as adapter side. I have few observations. Can someone please take a look and give suggestion?
=================
1. Adapters ports status is DOWN:
[root@RHEL7-R54-U5 ~]# connectx_port_config -s
--------------------------------
Port configuration for PCI device: 0000:1a:00.0 is:
eth
eth
[root@RHEL7-R54-U5 ~]# cat /sys/class/infiniband/mlx4_0/ports/1/state
1: DOWN
[root@RHEL7-R54-U5 ~]# cat /sys/class/infiniband/mlx4_0/ports/2/state
1: DOWN
[root@RHEL7-R54-U5 ~]# ibv_devinfo
hca_id: mlx4_0
transport: InfiniBand (0)
fw_ver: 2.33.5000
node_guid: f452:1403:00f4:f8c0
sys_image_guid: f452:1403:00f4:f8c3
vendor_id: 0x02c9
vendor_part_id: 4103
hw_ver: 0x0
board_id: MT_1090111019
phys_port_cnt: 2
port: 1
state: PORT_DOWN (1)
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
[root@RHEL7-R54-U5 ~]# ethtool enp26s0
Settings for enp26s0:
Supported ports: [ FIBRE ]
Supported link modes: 1000baseKX/Full
10000baseKX4/Full
40000baseCR4/Full
40000baseSR4/Full
Supported pause frame use: Symmetric Receive-only
Supports auto-negotiation: Yes
Advertised link modes: 40000baseCR4/Full
40000baseSR4/Full
Advertised pause frame use: Symmetric
Advertised auto-negotiation: Yes
Speed: Unknown!
Duplex: Unknown! (255)
Port: FIBRE
PHYAD: 0
Transceiver: internal
Auto-negotiation: off
Supports Wake-on: d
Wake-on: d
Current message level: 0x00000014 (20)
link ifdown
Link detected: no
2. Switch Port's admin state is enabled but operational state is down.
MLNX-R54-U15-40G [standalone: master] # show interfaces ethernet 1/1
Eth1/1
Admin state: Enabled
Operational state: Down
Description: N\A
Mac address: f4:52:14:65:55:06
MTU: 1500 bytes(Maximum packet size 1522 bytes)
Flow-control: receive on send on
Actual speed: 40 Gbps
Width reduction mode: disabled
Switchport mode: access
Last clearing of "show interface" counters : Never
60 seconds ingress rate: 0 bits/sec, 0 bytes/sec, 0 packets/sec
60 seconds egress rate: 0 bits/sec, 0 bytes/sec, 0 packets/sec
Rx
0 packets
0 unicast packets
0 multicast packets
0 broadcast packets
0 bytes
0 error packets
0 discard packets
Tx
0 packets
0 unicast packets
0 multicast packets
0 broadcast packets
0 bytes
0 discard packets
=================
Thanks!
Hi Komal,
Can you try the following:
ip link set enp26s0 up
Greetings,
I am working on a solution that includes interconnection between 2x SX1710 and 2x IBM EN4093R Flex System Switches using 40G QSFP+ for a distance of maximum 10 meters, utilizing SX1710 MLAG and IBM EN4093R switches VLAG
i am asking for the correct P/N of the QSFP+ transceiver to fit into SX1710, and the P/N of the fiber cable too, since the concept of DAC will not work in this case. Additionally, i want to confirm the support of PVRST+ on SX1710.
thank you
Hi,
I have the same problem but with English 2008 R2. I have tried and windows 7 but with no luck (same issues).
Does anybody know the resolution for this problem?
Kind regards,
Milan
Hi Eddie,
I tried setting up the link with the command you mentioned above. Still, port status is PORT_DOWN.
Hi Komal,
What is the switch software you are using?
run "show version concise"
what are the cable types and models?
run (from enable mode) "show interfaces ethernet transceiver"
Hi Eddie,
Thanks for looking into it.
I am using Fibre Optic Cable(speed upto 56GbE). I could see LED lightened in one port after I reconnected cables, but the other port(enp26s0d1) is still facing the issue. Port state is still down.
Here are the ans to your questions:
1.
MLNX-R54-U15-40G [standalone: master] # show version concise
SX_PPC_M460EX SX_3.3.5006 2014-05-20 12:19:44 ppc
2.
MLNX-R54-U15-40G [standalone: master] # show interfaces ethernet transceiver
List of transceiver details for all ports in the switch:
Port 1/1 state
identifier : QSFP+
cable/ module type : Optical cable/ module
ethernet speed and type: 56GigE
vendor : Mellanox
cable length : 10m
part number : MC220731V-010
revision : A1
serial number : DC321400013
Port 1/2 state
identifier : QSFP+
cable/ module type : Optical cable/ module
ethernet speed and type: 56GigE
vendor : Mellanox
cable length : 10m
part number : MC220731V-010
revision : A1
serial number : DC321400004
3. Port status:
[root@RHEL7-R54-U13 ~]# ibv_devinfo
hca_id: mlx4_0
transport: InfiniBand (0)
fw_ver: 2.33.5000
node_guid: f452:1403:00f4:f560
sys_image_guid: f452:1403:00f4:f563
vendor_id: 0x02c9
vendor_part_id: 4103
hw_ver: 0x0
board_id: MT_1090111019
phys_port_cnt: 2
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
[root@RHEL7-R54-U13 ~]#