[darookie]

Hello everybody,

I am doing a steady state simulation with a batch file.
After a 1800 iterations the solver did not give any new results out.
The solver did not stopped or crashed, so no .bak or .res file has been created.
My questions are:

1. Can I use any file like the mon, .out, .trk or .rg2 file to restart the simulation?
2. How can I stop a batch run, because the prompt window does not accept any input?

Thanks for your help,

Andrej

[nellio75]

Hello,

I think you cannot use the .mon files and so on. Moreover, if you use a windows OS, you can do "ctrl alt supp" and choose for the task manager, in the processes menu, order your processes by increasing CPU time and you will find your calculation (.exe). But if you kill your calculation there will be no results file.
With Linux use the command "kill -9 'number for your process calculation' ": you have the list of your processes with the command "top".

In other way your calculation seems to have an error if you can't visualise your results. Maybe you have to review your boundary conditions

I hope I answer your question.

Bye
nellio75

[darookie]

Thank you nellio. So i have to run it again. I am running the simulations under windows operation system.

I looked at the task manager and the programm solver-hpimpi.exe had just 0% of cpu use and just occupied disk space. How can this happen

[ghorrocks]

Can I use any file like the mon, .out, .trk or .rg2 file to restart the simulation?

No. The only files which can contain the information required for a restart are .def, .res, .trn, .bak.

How can I stop a batch run, because the prompt window does not accept any input?

The nicest way to stop a running simulation is to run the cfx5stop command. You can also do this by loading up the run in solver manager and pressing stop. This will nicely exit all tasks and shut down properly.

Sometimes you need to kill the job with the task manager (in Windows) or kill -9 (in Linux). Be aware that you will then have to manually remove the tasks off all nodes (if parallel) and probably some sub-tasks.

solver-hpimpi.exe had just 0% of cpu use

This means you are doing a hp-mpi simulation but it is locked up. Sometimes it unlocks itself and continues, sometimes not. It can lock for lots of reasons:
1) It is trying to grab more memory than is available on the machine so is starting to use swap (disk) RAM. This is very slow and means the CPU load fades to almost zero but the hard drive will be flogging itself.
2) Some problem in your parallel setup has stuffed the job.
3) Some problem in your simulation setup has stuffed the job.

[darookie]

Thank you very much Glenn, now I understand what has happened

[nellio75]

well done ghorrocks, I haven't thought about the file locking.

bye everybody

[smnaryal]

hello guys

Due to power failure, my pc shut down and the simulation was interrupted. Till now it has return 8 '.trn' files; and it is at 205th time step.

Is there any possibility that i could continue the simulation from the above timestep?

Suman

[ghorrocks]

If these trn files contain enough variables to initialise the run and include the mesh then yes. If not, then no.

Please start a new thread for a new question in future.

[smnaryal]

Thank u Glenn and your advice is appreciated. Will keep that in mind.