Learning Forum

[cct]

Hello,

I am trying to optimize a crash design in LS-OPT, but after running it, it shows the following error message.

ERROR: All analysis jobs failed.

       Double-click on the red Stage LED.

       Select a failed run and then "View log" to display the solver log.

       Common causes are:

            - Wrong command or name for the solver executable.

            - Job memory exceeded.

            - Termination statement (e.g. 'N o r m a l') missing at end of solver output.

                See GUI settings.

However, when I try to follow the steps to fix it, I double-click on the red stage LED, but there's no job shown in the window to select. Is there anything I missed?

[igandiko]

Please check job_log.solver file located in 1.1 directory of the stage. Do you see any errors, perhaps incorrect solver command? 

[cct]

I just checked the output file to see if it shows out some error.

In the upper side, it shows the key file is copied, but at the lower part, it shows the key file is not copied as shown in the following figures.

As well, the error message only shows the same one as I posted in the question.

[igandiko]

Please check job_log files located in 1.1 directory for errors. It seems ls-prepost command did not run correctly. Ls-opt runs job_command file to submit the run. You can try running this command file manually and see if it runs or not. 

[cct]

Just checked the job_log files located in 1.1 directory and found a license problem

and this error:

ERROR: No Binary Output Database (Binout) with root name "binout" found.

Current working directory is "/scratch/group/simu_anurag_ct/T005_09052022_tryaddpara_2/LS-Dyna/1.1".

Since I run LS-OPT in the school's cluster rather than a local computer, I need to ask how to run the command file manually and will take a try on it after getting an answer.

[cct]

In the job_log file, there's also a License problem.

I tried to ask for help from this email, and got the reply that post on ANSYS forum or submit a service request on the Ansys customer portal, but since I'm using school's system to run LS-OPT, I did not have the license info so I cannot do the second way.

[igandiko]

 

Since ls-dyna job did not run due to license problem, there is no binout file. Hence the second error. 

How are you running ls-dyna jobs? Locally (ls-dyna.exe i= input.k) or using some scripts to submit the run to cluster? If you can run ls-dyna locally, please make sure license environment variables are set correctly. These are,

setenv LSTC_LICENSE network

setenv LSTC_LICENSE_SERVER

You can also write a script, as shown below, and use this script in the place of solver command in the stage. 

#!//bin/bash/

export LSTC_LICENSE=network

export LSTC_LICENSE_SERVER=

/path/ls-dyna_solver i= DynaOpt.inp

If you are submitting to a cluster, use your ls-dyna job submitted script in the place of solver command and make sure you have “queuing” set to “simple” under remote options. 

A sample command could be,

qsub ../../submit_script.sh

../../ above is used to read the script located in working directory i.e. two levels up from job directory. 

[cct]

I opened the user interface in the cluster to run the LS-OPT, and the error message jumps out from it.

I also submit the run to cluster with the following commands, but it even did not run out the directory of LS-DYNA and LS-PREPOST as my first run (using ui to run the model).

[igandiko]

This looks like a script to run ls-opt in batch mode.

lsopt T005_003_xxxxx.lsopt

The above command runs lsopt project file, T005_xxxx.lsopt, in batch mode. When you open this project file in GUI, what do you have for ls-dyna solver command? You probably need a similar script but with last line (ls-dyna solver command) not commented out and you need to have i= DynaOpt.inp (instead of i= T005_parameters.k). 

[cct]

For ls-prepost command: /sw/hprc/sw/LS-PREPOST/4.8.9/lsprepost4.8_common/lspp48

For ls-dyna solver command: /sw/hprc/sw/LS-DYNA/R12.0.0/ls-dyna_smp_d

I changed the run file into , but the DynaOpt file seems created in the directory LS-Dyna>1.1 (and 1.2 to 1.4) after I ran in GUI. I tried to submit the run, but since it cannot detect the DynaOPt.inp in such directory, it shows the error "input file does not exist: DynaOpt.inp"

Do I need to move it into the main directory (same one as the run file)?

[igandiko]

Please also remove lsopt batch command from the script used to run ls-dyna. 

[cct]

I editted the run file (named run_lsopt) to be like the following figure.

Then, I changed the command in the stage of LS-Dyna as following.

Also have queuing set to simple under remote options. However, it still shows the same error and did not create new job_log file in directory 1.1.

[igandiko]

Did you clean old files before rerunning ls-opt? Tools > clean > clean all. 

You can also go to 1.1 directory, open a terminal and manually run sbatch ../../run_lsopt. See if ls-dyna runs this way. 

 

[cct]

I just manually delete the old files and rerun it. It seems work now. I'll let you know when it finished the job or fail.

[cct]

It seems work via GUI. Thank you so much!

Yet, it will take lots of time on running the optimization. Since I am not running in my local computer, I need to request a VNC for long time (and it will take time to get the access too) to open LS-Opt GUI. Could you please tell me the way to submit run job to cluster?