Ansys Remote Visualization Client Startup Guide

Using ANSYS Remote Visualization Client at Mines

This utilizes the GPU of your computer over CPU rendering on the HPC systems. For quick access to Ansys products use either wendian-ondemand.mines.edu or mio-ondemand.mines.edu for CPU rendering. The following application notes are adapted for use at Colorado School of Mines for our clusters, Wendian and Mio. For detailed information read the Fluent Workspaces User’s Guide, Chapter 1: Remote Visualization and Accessing Fluent Remotely, see Ansyshelp.ansys.com. Note that the Major and minor version and service pack numbers must match.

The purpose of using the ANSYS Fluent Remote Visualization Client is to access remotely running simulations without directly owning the process in case of the network disconnection. Using this application you can connect to a total of 6 running simulations and display the mesh, contour and vector plots, residual and monitor plots, and modify settings and send text commands to the TUI on the running Fluent (meshing or solver) server.

Step 1: Submit a Slurm Job

To get started create a SLURM submit script with the following in a directory with a Fluent case file. $HOME/scratch/$USER/fluent-cases/flremote-svr-start.submit

#!/bin/bash
#SLURM submit script filename: flremote-svr-start.submit
#SBATCH -t 00:30:00
#SBATCH -n1
#Opening 2-12 ports depending on host_node operation more than one server (2 minimum)
export REMOTING_PORTS=2678/portspan=2
# Loading Wendian module for Ansys version 2024R2
module load apps/ansys/242

# 3ddp for 3-Dimension double precision; -t1 for 1 cpu; -g for enabling graphics; -driver null for no graphical interface
# -sifile= for enabling the fluent server and giving a filename to store IP and PASSWORD
# use -i journal_of_TUI_commands.jou
# Filename starts with the variable SLURM_JOBID for uniqueness 

fluent 3ddp -t1 -g -driver null -sifile=./"$SLURM_JOBID"_fluent_server_info.txt

Now, submit this to the SLURM scheduler

[joeuser@wendian001.mines.edu]$ sbatch flremote-svr-start.submit
Submitted batch job 6033061

Once the Fluent job starts the file with the SLURM_JOBID_fluent_server_info.txt is created with the connection information, the Fluent cleanup script (can be run by executing sh FILENAME at the command prompt), a transcript files, and the slurm-SLURM_JOBID output file. The connection information file identifies the IP address and password needed to create the network port tunnel from the client machine to this node.

For example, this job now has the follow files in the working directory.

[joeuser@wendian001.mines.ed]$ ls 
6033061_fluent_server_info.txt  cleanup-fluent-node136-20180.sh  fluent-20210216-154942-19982.trn  slurm-6033061.out
catalytic_converter.cas         flremote-svr-start.submit

As an alternative to an HPC running server. You may test running this from a Fluent GUI (Linux or Windows) by selecting “File” -> “Start/Stop Server” or run the TUI “server/start-server”. You can connect to local running copies of the server on the same machine to practice setting up commands or running scripts.

Step 1: Alternative Startup using a Journal File

Submit a Fluent job using predefined commands

Fluent executes commands through the TUI (Text User Interface). The format is a command followed by white space, comma, or answer to the prompted question. For a detailed list see the Ansys Help documentation for Fluent -> Fluent Text Command List -> Solution Mode. General navigation from a the fluent console is a directory structure. Press ‘enter/return’ will list available entries that are either commands or directories. Pressing ‘q’ will take you back one, and entering a command in the shortest unique letters will begin the prompts for options regarding that command. For example, ‘define/models/viscous/ke-realizable?’ is a simple yes or no, but ‘/define/boundary-conditions/velocity-inlet’ requires further information such as which surface(s), velocity, turbulent parameters, temperature, species, etc.

The a simple journal file may contain a read case, initialize or read data, iterate, save case and data, and exit. However, for our purposes here starting a visualization client server and reading a case is sufficient.

The file read_case.jou contains these two lines:

;Journal file to start the visualization client server and read a case
server/start-client
file/read-case myFluentCase.cas

Then the command in the batch script to start fluent changes to call this journal file. Using multiple processors and nodes the variables $cpus and $SLURM_JOB_ID.nodes are created in the script See Fluent using Multiple Nodes. So the last line becomes:

fluent 3ddp -t$cpus -cnf=$SLURM_JOB_ID.nodes -gu -i read_case.jou

Step 2: Get the connection infomation and start a tunnel

Output the contents of the fluent server file: SLURM_JOBID_fluent_server_info.txt or use the Ondemand “Files” tab to view the text file.

[joeuser@wendian001.mines.edu]$ cat 6033061_fluent_server_info.txt
172.18.10.78:2678
229zknbn

Edit or copy the content to a file on your local machine and change the IP address to “localhost”, for example on my desktop this file looks like this now: C:\Users\joeuser\Desktop\6033061_fluent_server_info.txt

localhost:2678
229zknbn

Starting a Tunnel to the compute node host

Opening a Terminal (Linux) or Windows PowerShell and run the ssh command with the flag -L with the connection information.

Windows PowerShell
Copyright (C) Microsoft Corporation. All rights reserved.

Install the latest PowerShell for new features and improvements! https://aka.ms/PSWindows

PS Z:\> ssh -L 8001:c078:2678 joeuser@wendian.mines.edu

The first number 8001is your choice, the compute host name is c078, and the second number 2678 is the port number gave in the submit script.

Step 3: Start your local client and connect

Start the “Remote Visualization Client 2024 R2” App from the windows start menu. Locate the “Server info Filename” box under “Properties - RemoteSession 1” and click in the box. Select the file on your Desktop and the client will connect through your localhost port through the ‘ssh’ encripted tunnel to the compute node.

From here you can view the mesh, watch the simulation run, send commands to the server, and disconnect knowing the job is still running.