Hi,

I'm running distributed parallel jobs on a Linux-cluster (2 dual proc nodes and 1 master). Sometimes my job exits suddenly with the following message: The CFX-5 Solver exited with return code 1. No results file has been created.

What is return code 1? It's very annoying, because sometimes this happens after a lot of calculation hours, and because there is no results file, I cannot restart the job.

Regards, Jaap

Me too. I met this problem so many times. I think it's the software's problem.

Hannah

Hi Jaap

As a matter of course, I always run the CFX-5 Solver with Backup Results switched on (usually every 20 to 50 iterations), which can be used to restart the Solver should the run fail.

Regards,

Sharks

Of course you should turn on the backup option. However, when using this option, you lose all your RMS residual data of the previous steps AND I cannot determine at which nodes my maximal RMS residuals have taken place. Do you have a solution for this problem?

Regards,

Jaap

Thank you for your response. On which platform are you running CFX 5?

Regards Jaap

Just set turn on equation residuals output in CFX-Build. You could look at them in CFX-Post.

As for your original problem, is there any message in the output file in addition to the return code 1 exit message. Maybe the solver is crashing, or something, did you take the time to look at the output file and figure it out? It might not be the software either, sometimes a poorly setup and maintained Linux cluster (eg: you just slapped together a bunch of cheap PCs) will create problems. Maybe you need some new network cards, or new motherboards or something?

Neale.