Inconsistent errors

General Usage
1

I’m running the same script multiple times, around 1000 times, and I get inconsistent errors.
The errors I randomly get are:
1) “ERROR: LoadError: syntax: invalid character “” near column 1”
2) “ERROR: LoadError: DimensionMismatch(“new dimensions (300, 1885) must be consistent with array size 2035800”)”
3) “Error encountered while load FileIO.File{FileIO.DataFormat{:JLD}, String}(”/storage/ph_kafri/ranya/OpenBoundaries/Data/Avg/Rho10T0.555V1.0Dx1.0Dy1.0G1.0numfile0.jld")."

I would have guessed that all these errors should always occur, but they do not, just randomly whenever they want to.
I can say that the scripts run simultaneously and that my only lead so far regards global and local variables.
I can add any code you would like, but the code itself is 500 lines…
I would really appreciate any kind of help here as this is very frustrating.

2

Do you use threads? If you don’t start Julia that way you should always (or never) be able to load the same file. I guess you should always be able to load even with using threads, so that brings me to the next question:

Is that file for sure static? Or are you making it with your program and then reading it? Then why isn’t it always the same, what’s non-deterministic about the process, what inputs go into making that file? Again could be about threads, or do you use rand?

I would try to run like:

julia --check-bounds=no

and skip -t

3

Thanks for the reply:)
I’m using multiple cores that execute the same script, so I would say that I’m not using threads if I’m correct.

I’m positively sure that the file is static, the program should only read it and never write into it, and as far as I understand the process should be completely deterministic, at least for that part of the script.
I am using rand in my code, but it is used for running the simulation itself and not for loading any file.
The file I’m reading is composed of 2 large arrays (around 10^7 each) and 3 integers, and I’m using load in order to read it.

Can you explain what is the meaning of adding --check-bounds=no when running a Julia script? and also what did you mean by skip -t?

Could it also relate to the way Julia is running the script? is it guaranteed that it goes line by line or can it try to run multiple lines simultaneously to increase performance?
Thanks in advance.