[chris_sev]

Hallo,

When I run nonNewtonianIcoFoam in parallel transient results are not beeing written on nodes other than master.
I have never experienced this in case of other solvers. I am sure that I didn't messed with settings because when I use other solver for the same simulation and number of processors everything works fine.

Is that bug?

[mattijs]

I cannot repeat this with the offsetCylinder tutorial.

All the field writing is done in the same way regardless of the solver so it is unlikely it is a problem of nonNewtonianIcoFoam itself. Is your problem only for one case or also for the tutorial case?

[chris_sev]

Hi Mattijs,

Thanks a lot for response.
With tutorial case (offsetCylinder) it's the same. I've just checked this, used 2computers with 2 processors on each.
I got transient results only on my master machine (processor0 and processor1).

I am using standard nonNewtonianIcoFoam that comes with installations package. I didn't played with the code (apart from implementation of time step adjustment).
I am using 1.5 version.

Do you have any ideas/hints ?

Thanks in advance,
Chris

[mattijs]

Run all (4?) your processes on your local computer and check. Maybe your second computer does not allow writing to the disk.

Switch on debug flag for objectRegistry. You'll get messages from all processors of the form

objectRegistry::write() : Considering writing object ...

[chris_sev]

I am not very familiar with parallel run in OF so i don't realy know what does it mean and how to manage this:

Originally Posted by mattijs

Switch on debug flag for objectRegistry. You'll get messages from all processors of the form

objectRegistry::write() : Considering writing object ...

But anyway thanks for help. I am sure it's helpfull, so now I will search through the forum trying to figure out what's the meaning of above

I am wondering also how it's possible that other computer doesn't allow to write data when use nonNewtonianIcoFoam but when use rasInterFoam everything is fine :/

Chris