[mvapich-discuss] MVAPICH2-GDR 2.2 run issue

[Na Zhang]

Hi Sourav,

Thanks for getting back to me. I checked. Both hosts have firewall disabled
and ssh free among each other.

Any other possible issues for causing the errors?

Na

On Mon, Oct 16, 2017 at 8:54 PM, Sourav Chakraborty <
chakraborty.52 at buckeyemail.osu.edu> wrote:

> Hi Na,
>
> This looks like an issue with your firewall or ssh settings. Are you able
> to ssh into the nodes without entering a password?
>
> Thanks,
> Sourav
>
>
> On Mon, Oct 16, 2017 at 4:57 PM, Na Zhang <nkuzhangna at gmail.com> wrote:
>
>> Hi MVAPICHdevelopers and users:
>>
>> I'm trying to install MVAPICH2-GDR 2.2 and test whether GPU Direct works
>> in my environment.
>>
>> I installed MVAPICH2-GDR 2.2  (mvapich2-gdr-2.2-4.cuda8.0.m
>> ofed4.0.gnu4.8.5.el7.centos.x86_64.rpm) and set up environmental
>> variables and used the following command to test:
>>
>> export MV2_PATH=/opt/mvapich2/gdr/no-mcast/2.2/cuda8.0/mpirun/gnu4.8.5
>> export MV2_GPUDIRECT_GDRCOPY_LIB=/opt/mvapich2/gdr/no-mcast/2.2/cud
>> a8.0/mpirun/gnu4.8.5/lib64
>> export MV2_USE_CUDA=1
>> export MV2_IBA_HCA=mlx4_0
>>
>> $MV2_PATH/bin/mpirun_rsh -debug -np 2 host1 host2
>> LD_PRELOAD=$MV2_PATH/lib64/libmpi.so $MV2_PATH/libexec/osu-micro-benchmarks/get_local_rank
>> $MV2_PATH/libexec/osu-micro-benchmarks/osu_latency D D
>>
>> The debug message shows:
>>
>> /usr/bin/xterm: Xt error: Can't open display:
>> /usr/bin/xterm: DISPLAY is not set
>> [bitfusion-master:mpirun_rsh][child_handler] Error in init phase,
>> aborting! (0/2 mpispawn connections)
>> /usr/bin/xterm: Xt error: Can't open display:
>> /usr/bin/xterm: DISPLAY is not set
>>
>> I'm using Ubuntu 16.04, CUDA 8.0 and OFED 4.1-1.0.2. Each host has one
>> NVIDIA P100 and one Mellanox Connect X-3 card (Infiniband mode). And I have
>> Mellanox's nv_peer_mem module installed on the two hosts.
>>
>> Any suggestions are appreciated. Thanks.
>>
>> Best,
>> Na
>>
>>
>>
>> _______________________________________________
>> mvapich-discuss mailing list
>> mvapich-discuss at cse.ohio-state.edu
>> http://mailman.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20171018/503c0a4b/attachment.html>