Remote Processing Setup

dlh
dlh New Altair Community Member
edited October 2020 in Community Q&A

Greetings,

 

I'm having a little trouble with simulating jobs remotely, specifically using the front-end on Windows pre-process and then remote solving on a Linux machine. I have setup ssh with passwordless login and confirmed that it works between my two machines. Next, I try launching from the gui with the appropriate Component launch options set on the FEKO Solver tab, but no joy. I then open a Feko terminal and put the particulars on the command line as follows: 

 C:\Users\user1\projects\fy19\antenna_lib\ant01>runfeko ant05l_matched2 --remote-host 192.168.1.11 -np 12 --machines-file C:\Users\user1\AppData\Roaming\feko\2018.0_caf82d78\machines.feko --parallel-authenticate default

and get the following response in the terminal: 

  C:\Users\user1\projects\fy19\antenna_lib\ant01>runfeko ant05l_matched2 --remot e-host 192.168.1.11 -np 12 --machines-file C:\Users\user1\AppData\Roaming\ feko\2018.0_caf82d78\machines.feko --parallel-authenticate default  Altair Feko - RUNFEKO Version 2018.2.1-12 from 2018-10-19           Copyright (c) 1999-2014 Altair Development S.A. (Pty) Ltd           Copyright (c) 2014-2018 Altair Engineering, Inc.  NOTE: Running PREFEKO locally since using remote launching  Altair Feko - PREFEKO Version 2018.2.1-63 from 2018-12-07           Copyright (c) 1999-2014 Altair Development S.A. (Pty) Ltd           Copyright (c) 2014-2018 Altair Engineering, Inc.   Creating file ant05l_matched2.fek, please be patient ... File ant05l_matched2.fek has been created:       1 metallic wire segments in free space (of which 1 are curvilinear)     388 metallic triangles in free space   Altair Feko - ADAPTFEKO Version 2018.2.1-15 from 2018-10-12           Copyright (c) 1999-2014 Altair Development S.A. (Pty) Ltd           Copyright (c) 2014-2018 Altair Engineering, Inc.   Altair Feko - PREFEKO Version 2018.2.1-63 from 2018-12-07           Copyright (c) 1999-2014 Altair Development S.A. (Pty) Ltd           Copyright (c) 2014-2018 Altair Engineering, Inc.   Creating file ant05l_matched2.fek, please be patient ... File ant05l_matched2.fek has been created:       1 metallic wire segments in free space (of which 1 are curvilinear)     388 metallic triangles in free space   --------------------------------------------------------------- ADAPTFEKO frequency analysis band no. 1 of 1 ADAPTFEKO analysis no. 1 (max. 75) at frequency  5.00000e+07 Hz           Completed subintervals:  0.0% ---------------------------------------------------------------   Altair Feko - RUNFEKO Version 2018.2.1-12 from 2018-10-19           Copyright (c) 1999-2014 Altair Development S.A. (Pty) Ltd           Copyright (c) 2014-2018 Altair Engineering, Inc.  NOTE: Running PREFEKO since file 'ant05l_matched2_fr_1_ada_1.fek' not found  Altair Feko - PREFEKO Version 2018.2.1-63 from 2018-12-07           Copyright (c) 1999-2014 Altair Development S.A. (Pty) Ltd           Copyright (c) 2014-2018 Altair Engineering, Inc.   Creating file ant05l_matched2_fr_1_ada_1.fek, please be patient ... File ant05l_matched2_fr_1_ada_1.fek has been created:       1 metallic wire segments in free space (of which 1 are curvilinear)     388 metallic triangles in free space  Launching of the job remotely on the host '192.168.1.11' Copying files to remote host ...   Altair Feko - RUNFEKO Version 2018.2.1-12 from 2018-10-19           Copyright (c) 1999-2014 Altair Development S.A. (Pty) Ltd           Copyright (c) 2014-2018 Altair Engineering, Inc.  Host key verification failed. Warning: List of hostnames not found. - Temporary directories on nodes not delet ed.  ERROR  20032:   Error opening the source file 'ant05l_matched2_fr_1_ada_1.out' for copying.   C:\Users\user1\projects\fy19\antenna_lib\ant01>

Any suggestions on how to overcome this error and get this model solved remotely?

 

Thanks.

 

v/r,

 

DLH

Tagged:

Answers

  • Mel
    Mel Altair Community Member
    edited February 2019

    Hi dlh

     

    First try remote launching without using parallel. So please remove this part from your command 

     -np 12 --machines-file C:\Users\user1\AppData\Roaming\feko\2018.0_caf82d78\machines.feko --parallel-authenticate default

    If the above works, then there is a problem with parallel on the remote machine.

    Confirm, by opening a terminal and logging in on the remote machine, that you can launch a parallel FEKO job there.

     

    If you still struggle, add the debug output here. You can add debug output by adding -d after runfeko, e.g.

    runfeko mymodel -d  --remote-host 192.168.1.11

  • dlh
    dlh New Altair Community Member
    edited February 2019

    Hi mel,

     

    I did as you suggested and the simulation now runs remotely, but only on 1 core. So, now how does one take advantage of multiple cores on the remote system? If I login to the remote system, and launch runfeko from the command line there, or even the gui, I can take advantage of multiple cores with no issues so parallel processing does work on the remote system.

     

    Thanks!

  • Mel
    Mel Altair Community Member
    edited February 2019

    Hi dlh

     

    Have you tried using the same machines file for the parallel test when running while logged into a terminal directly on the remote host?

    Your previous output showed ' 

     Host key verification failed.

    This still shows something's up with the ssh setup.

     

    Please (1)  send the debug output, but please use an example without continuous frequency (generates too much unnecessary output) - you can use a model from ExampleGuide model A02.

    And (2) send the machines file as well.

     

    Thanks

     

     

     

     

Categories