Open ClaraPicoron opened 10 months ago
This is their Python configuration:
Hello, has there been any progress on this issue ? Thank you.
I'm not aware of the full context of this issue, but there have been Slurm/HPC improvements in PyFluent:
As of version 0.18.2, if you manually launch Fluent in a Slurm environment using a command such as
fluent 3ddp -scheduler=slurm -scheduler_queue=cdc01 -sifile=server-info.txt -t28
You will be able to connect PyFluent to that Fluent instance (after the Slurm job starts running), using for example:
solver = pyfluent.connect_to_fluent(
server_info_filepath="server-info.txt",
cleanup_on_exit=True,
)
cleanup_on_exit = True
used in this example so that when solver.exit()
is called in PyFluent, Fluent will exit and release the HPC resources.
The improvements coming in PyFluent 0.19 will make it easier to launch Fluent in a Slurm queue using PyFluent, as shown in the link I shared above.
PyFluent 0.19 was released yesterday. Any updates on the way of launching Fluent in a Slurm queue?
@christospliakos You should be able to launch it as documented. I've replied in your issue #2351.
@ClaraPicoron The launch option does not look right to me if that's run within the SLURM system. start_instance=0
is passed which means that an instance must already be running. Can they try:
solver = pyfluent.launch_fluent(precision="double", mode="solver" )
and then does the script run? They should not need to pass the server info file argument at all unless they have previously started the job.
Also, please have a look at my post from #1548. It's still not clear to me what the customer is trying to do.
@dnwillia-work , the customer tested this:
Unfortunately the serverinfo file is still not created with this error:
Originally posted by @ClaraPicoron in https://github.com/ansys/pyfluent/issues/1548#issuecomment-1782981923