Quantcast
Channel: Mellanox Interconnect Community: Message List
Viewing all 6226 articles
Browse latest View live

Re: HowTo Uninstall ONIE/Cumulus and Replace it with MLNX-OS

$
0
0

Hi Fabrice,

 

 

see below:

 

 

My questions are:

 

1/ I would have to regularly reset three of them in order to go back to a ''clean'' environnement.

     I have read many things about it, but apparently i will need to ''reactivate'' the Cumulus license each time i reset. Is it right ? If so, is it always the same key or do i need to call Mellanox each time ?

 

 

Answer - Same cumulus license file will work every time on the same switch

 

 

 

2/ On the one that will be used for myself. I would like to uninstall ONIE/Cumulus and replace it with MLNX-OS. Cause i am mainly familiared with it.

     Is it possible ? In my opinion, reading this article : HowTo Uninstall MLNX-OS and Install ONIE on Mellanox Switches , it looks...

     Do I need to purchase a licence of MLNX-OS ? Or will it be integrated (''offered'') with the switch, as this is a Ethernet switch ?

 

Answer - Please contact support@mellanox.com for this support contact question - please mention the Switch PArt number you are planning to buy


Re: HowTo Uninstall ONIE/Cumulus and Replace it with MLNX-OS

$
0
0

Thank you very much for your quick response and your professionalism!
Best regards.

Re: HowTo Uninstall ONIE/Cumulus and Replace it with MLNX-OS

$
0
0

Hi,

As you requested me, i've emailed the Mellanox support regarding my second question.

I've sending them the link to this community thread in order they can also have the context of my questions...

 

The only thing they say me is :

" According to our CRM records we could not locate any valid contract for you, please use our community portal or contact our
contracts department If we mis-identified your account, please send us your account name and contract number in order to proceed."

 

Sorry, but they are not professionnal at all, cause it's not a support question but a pre-sale question, although it deals with the "technical" field of your product.

 

In conclusion, who can, or WANTS, to respond to my second question, without having to first pay a $$$$ contract, before possibly buying ??!!

 

Sincerely yours.

Re: HowTo Uninstall ONIE/Cumulus and Replace it with MLNX-OS

$
0
0

Hi Fabrice,

 

someone will respond through the case.

Understanding vma multi tread tuning and what are rings?

$
0
0

Hello community.

I got some questions.

We are using mellanox with libvma for getting Market Data from exchanges.

For example we got 4 igmp v3 subscriptions.

I tried to use vma with isolcpus and cpusets for it but it always using first core from cpuset lists (for example 0-4).

So the questions are:

1. how to correct understand what is rings? (I mean ring allocations)

2. What is the best practice for using vma as multi treads? I mean should it be on the same core as listening socket or not?

3. How does vma works with data? Does it send all by rdma to ram, or it using multiply treads for parsing data?

4. How we can tell vma use different cores for different subscriptions?

5. Does vma sending data to other cores by L3 cpu cache?

Re: ConnectX-3 Pro VPI/EN Receive VXLAN Offloading

$
0
0

In order to validate receive VXLAN offload, you can configure the system by using 'ethtool' and settingg rx-checksumming and GRO on receive side, run the traffic and observe that receive packet size is bigger than MTU using wireshark/tcpdump  on receiver side.

Re: mlx5 direct verbs example

Re: Scalability issue for multiple clients

$
0
0

It might be useful if you can transform your test description to something that uses ib_read_bw/ib_write_bw or iperf (if you are using TCP) and show what is the output.

Can you see any drops in 'ethtool -S' output or device statistic ( ifconfig, ip)?

Does the sender use 1 CPU when writing to two different clients? Or, in other words, does he use the same thread?

You might check what is the output of 'mlnx_perf' command, however you need Mellanox OFED installed on the host.


Re: Mellanox ConnectX-5 adapter speed 650MB/s only

Re: Problema con Tarjeta infiniband Qlogic 4x QDR IB PCI-E G2 HCA

$
0
0

Thanx for the info but still i cant find any driver i also talk to HP and tehy dindt gave me an answer but still thank you!

Re: Mac address on routed ports belonging to separate VRFs are same

$
0
0

Is this an incomplete configuration? I don't see IP addresses on these interfaces. Is IP routing configured, and vrf route distinguisher. The interfaces shown are all down. Do the mac addresses displayed change once they are UP?

Re: Where's the procedure of packing network protocol header in RoCE v2?

$
0
0

It is expected behavior.

HCA can support up to 16 millions QPs. But there are only up to 64K UDP port number, and some UDP port was reserved for some specific application. So it is not 1:1 mapping between QP and src UDP port, but N:1 mapping. Always, a couple/group of QPs share one src UDP port. The main purpose of using scrambled src UDP port is for load balance in network. N:1 mapping works for that.

Error MessageSystem.QueryException: unexpected token: '('

$
0
0

Please could someone help me

Error MessageSystem.QueryException: unexpected token: '('

Stack TraceClass.batchAccountUpdate.start: line 4, column 1

 

I get this error message when I am trying to run the test class for this batch

 

 

01global classbatchAccountUpdate implementsDatabase.Batchable<sObject>{

 

 

02    global Database.QueryLocator start(Database.BatchableContext BC) {

 

 

03        Stringquery='SELECT id, Name FROM Opportunity WHERE Createddate =:system.today() AND Status <> \'closed\' AND Has_IBM_Product__c = False AND Id IN : (SELECT OpportunityId FROM OpportunityLineItem WHERE pricebookentry.product2.family =\'XYZ\') limit 10';

 

 

04       

 

 

05        returnDatabase.getQueryLocator(query);

 

 

06    }

 

 

07  

 

 

08    global voidexecute(Database.BatchableContext BC, List<Opportunity>scope) {

 

 

09         for(Opportunity Opp : scope)

 

 

10         {

 

 

11             Opp.Has_ABC_Product__c=True;           

 

 

12         }

 

 

13         updatescope;

 

 

14    }  

 

 

15   

 

 

16    global voidfinish(Database.BatchableContext BC) {

 

 

17    }

 

 

18

}

 

Thank in Advance

New to InfiniBand, so we are unable to configure to communicate between server and switch

$
0
0

Dear Member

 

I am new to InfiniBand I have a problem I have  7 server and one Mellanox switch we need to configure ethernet over InfiniBand

As I am new to InfiniBand technologies, so please help us to sort out the issues as below.

 

Installation steps

 

  • We had 7 no of servers with one of ConnectX-3 Single Port Card in each server.
  • I have one number msx6005f-1bfs switch
  • In each server operating system is centos 7.4
  • I am mark in checkbox to install InfiniBand supporting software during OS installation.

Issue we are facing

 

  • All server connected to InfiniBand switch but only number one port orange led is glowing in switch end
  • We are unable to configure IB over Ethernet

 

Please provide the steps to configure the InfiniBand Switch & communicate between Server & Switch for High Performance Computing(HPC).

More information about Innova Flex

$
0
0

Hello!

 

We are currently looking at Innova Flex cards to implement our IDS. We wanted some help on the following questions:

 

1) We understand that the FPGA on this card is "bump in the wire". We also understand the input to the FPGA is the network packet coming from the switch. What is the output? Is it also a network packet which will then go the adapter (and then to the server where the adapter is connected)?

2) Can we send the packet out directly from the FPGA? For example, we receive the packet from the switch, perform our processing, and then directly send the packet back to switch without sending it to the adapter (and the server)?

3) How do we program the FPGA? We have some familiarity in programming Xilinx FPGAs using Vivado. Can we use the same tool to program the FPGA?

4) Does this FPGA connect to the main DRAM (on the server), or can only use the on-board (2GB) DRAM?

 

P.S. The Innova Flex card looks awesome!

 

Thank you for your help in advance!


Re: Mellanox ConnectX-5 adapter speed 650MB/s only

$
0
0

Thank you. I tried NTttcp with different send buffer size and datagram size. I was able to get 4.8 GBytes speed on 40Gbps link.

mlnx-fw-updater-4.1-1.0.2.0 fails

$
0
0

Hi Mellanox community,

 

System: Dell PowerEdge C6320p

OS: CentOS 7.3

IB Controller: Mellanox Technologies MT27700 Family [ConnectX-4]

OFED: MLNX_OFED_LINUX-4.1-1.0.2.0-rhel7.3-x86_64

 

I'm having a problem on installing MLNX_OFED_LINUX-4.1-1.0.2.0-rhel7.3-x86_64 on Dell PowerEdge C6320p. The problem is that the installation of mlnx-fw-updater-4.1-1.0.2.0 fails with the following error message:

 

The firmware for this device is not distributed inside Mellanox driver: 03:00.0 (PSID: DEL2180110032)

To obtain firmware for this device, please contact your HW vendor.

EXIT_STATUS: 2

 

As the message suggested, I went to the HW vendor(DELL)'s website, and found this download link, which is redirected to Mellanox's site where I got the MLNX_OFED_LINUX-4.1-1.0.2.0-rhel7.3-x86_64. I asked Dell Tech Support and was told to try an older version, 4.0-1.0.1.0, instead, and then ended up getting the same error. That's why I'm here, asking the question to the community, hoping someone knows the solution.

 

I would appreciate any feedback.

 

Thanks,

Re: New to InfiniBand, so we are unable to configure to communicate between server and switch

$
0
0

HI Team,

 

i am configure infiniband OFED Driver after install

 

ibstat

CA 'mlx4_0'

        CA type: MT4099

        Number of ports: 1

        Firmware version: 2.36.5000

        Hardware version: 1

        Node GUID: 0xe41d2d03004e8520

        System image GUID: 0xe41d2d03004e8523

        Port 1:

                State: Down

                Physical state: Polling

                Rate: 10

                Base lid: 0

                LMC: 0

                SM lid: 0

                Capability mask: 0x0251486a

                Port GUID: 0xe41d2d03004e8521

                Link layer: InfiniBand

 

and also find open sm log

 

cat /var/log/opensm.log

Oct 26 12:55:55 152889 [31773740] 0x03 -> OpenSM 4.9.0.MLNX20170607.280b8f7

OpenSM 4.9.0.MLNX20170607.280b8f7

 

Oct 26 12:55:55 152971 [31773740] 0x80 -> OpenSM 4.9.0.MLNX20170607.280b8f7

 

No local ports detected!

Oct 26 12:55:55 199919 [31773740] 0x02 -> osm_vendor_init: 1000 pending umads specified

Oct 26 12:55:55 199989 [31773740] 0x02 -> osm_vendor_init: 1000 pending umads specified

Oct 26 13:02:00 186218 [C18E4740] 0x03 -> OpenSM 4.9.0.MLNX20170607.280b8f7

OpenSM 4.9.0.MLNX20170607.280b8f7

 

Oct 26 13:02:00 195147 [C18E4740] 0x80 -> OpenSM 4.9.0.MLNX20170607.280b8f7

Using default GUID 0xe41d2d03004e8521

Entering DISCOVERING state

 

Oct 26 13:02:00 205192 [C18E4740] 0x02 -> osm_vendor_init: 1000 pending umads specified

Oct 26 13:02:00 205269 [C18E4740] 0x02 -> osm_vendor_init: 1000 pending umads specified

Oct 26 13:02:00 251984 [C18E4740] 0x80 -> Entering DISCOVERING state

SM port is down

 

Oct 26 13:02:00 252197 [C18E4740] 0x02 -> osm_vendor_bind: Mgmt class 0x81 binding to port GUID 0xe41d2d03004e8521

Oct 26 13:02:00 295259 [C18E4740] 0x02 -> osm_vendor_bind: Mgmt class 0x03 binding to port GUID 0xe41d2d03004e8521

Oct 26 13:02:00 338453 [C18E4740] 0x02 -> osm_vendor_bind: Mgmt class 0x04 binding to port GUID 0xe41d2d03004e8521

Oct 26 13:02:00 338491 [C18E4740] 0x02 -> osm_vendor_bind: Mgmt class 0x21 binding to port GUID 0xe41d2d03004e8521

Oct 26 13:02:00 338525 [C18E4740] 0x02 -> osm_opensm_bind: Setting IS_SM on port 0xe41d2d03004e8521

Oct 26 13:02:00 340619 [A605A700] 0x80 -> SM port is down

OpenSM: Got signal 15 - exiting...

Exiting SM

 

pleas suggest

General Switch Questions: Traffic Control?

$
0
0

I don't know if this is the appropriate place to discuss general Infiniband questions, so please feel free to point me to the correct forum.

 

I'm working on an real-time application where network communication is crucial.  How do Infiniband switches manage traffic control?  If multiple hosts (A and B) try to send large amounts of data (gigabytes) to a single host (C), what happens?  I assume the messages from A and B are broken down into smaller packets, and then those packets arrive at C in a non-deterministic interleaved order.  That is how TCP/IP would work.  Does Infiniband do this any differently?  Is there a way to prioritize A over B so that the data from A arrives first?

 

Does the Subnet Manager accomplish this in the Infiniband architecture?  Or does the Subnet Manager do something completely different?

 

Also, consider a different situation where C sends a large message to both A and B.  Can this message be once sent to both A and B simultaneously?  Or does the message have to be sent twice: once to A and once to B?

 

Jim

Re: RDMA read failing with Remote Invalid Request Error

$
0
0

I'm having a similar problem with the RHEL Inbox drivers and NFS over RDMA between some newly installed machines, getting Local Length Errors. While I don't have a solution for you specifically, I'm wondering, how did you do the validation you mention (qp_access flags, buffer sizes, etc.)? I'm wondering if any of the things you're mentioning might help me with my problem.

Viewing all 6226 articles
Browse latest View live


<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>