[Sept24]

Hi All,
My udf interpreted well and my simulation runs smoothly on my PC. However, when I try to run the same thing on supercomputer (batch system), it doesn’t work. Is it correct that to run on batch, I set up my problem as usual, interpreted and hooked the udf, initialize and save project (one step less if I want to run on my PC, where after initialization, click run calculation>calculate). Here is my inputfile (connected with PBS submission file at the line: fluent 3d -g -t 4 -i < FlowTestInput.txt > Solution.out)

FlowTestInput.txt :
/define/user-defined/interpreted-functions “UDFSource.c” “cpp” 10000 no
/file/read-case FlowTest.cas.gz
/solve/initialize/initialize-flow
/solve/set/time-step 2
/solve/dual-time-iterate 100 1000
/file/write-data FlowTest.dat.gz
exit
yes

Here are the errors that appeared (cut and paste from .o and .out file)

Error: chip-exec: function "source_C" not found.

Error: UDFSourceTerm.c: line 4: syntax error.
cpp: output pipe has been closed

Error: File E:AC20UDFSourceTerm.c doesn't exist!
Error Object: #f

Primitive Error at Node 2: received Segmentation Violation
Primitive Error at Node 3: received Segmentation Violation
Primitive Error at Node 0: received Segmentation Violation
Primitive Error at Node 1: received Segmentation Violation

Error: Compute node encountered error
Error Object: #f
Primitive Error at Node 2: Update_Time_Level: invalid data
Primitive Error at Node 1: Update_Time_Level: invalid data
Primitive Error at Node 3: Update_Time_Level: invalid data
Primitive Error at Node 0: Update_Time_Level: invalid data
Error: Compute node encountered error
Error Object: #f
Primitive Error at Node 1: Write_Data_Section: uninitialized flow field
Primitive Error at Node 3: Write_Data_Section: uninitialized flow field
Primitive Error at Node 0: Write_Data_Section: uninitialized flow field
Primitive Error at Node 2: Write_Data_Section: uninitialized flow field
56.4%
Error: Compute node encountered error
Error Object: #f
Error: Error writing "| gzip -2cfv > AC20.dat.gz".
Error Object: #f

Please tell me what are the extra steps that should be taken care of when submitting job to batch system involving UDF? Also, please advise me on what causing the errors.
Note: Processing options: serial

Thank you in advance.

[Sept24]

With some helps from my friends, problem solved! The problem occurred because my file was originally created on windows PC whereas the supercomputer system use unix machine. To avoid all the errors, transfer/upload all necessary files to the supercomputer system as usual, but before submitting the job, convert the udf file from dos to unix. To do that, on the unix command line on the supercomputer system type (let say the udf file name is UDFSource.c)

dos2unix UDFSource.c

Good luck everyone

[asal]

Hello.

I try the same command
dos2unix UDFSource.c

But I got this error:
bash: dos2unix: command not found
what should I do?

[roi247]

Originally Posted by asal

Hello.

I try the same command
dos2unix UDFSource.c

But I got this error:
bash: dos2unix: command not found
what should I do?

First trial: I couldn't use ""sign in the file
second trial: dos2unix command couldnt be found
third trial: I couldn't write dos2unix directly on the super computer