Simulation Error Feedback via PyRocky not supported?

🔍 Before submitting the issue

• I have searched among the existing issues
• I am using a Python virtual environment

🐞 Description of the bug

Exception has occurred: SerializeError
unsupported serialized class: sci20.simulator.interfaces.SimulationError
File "XYZ/xyz.py", line 56, in start_sim
self.study.StartSimulation(non_blocking=True)
Pyro5.errors.SerializeError: unsupported serialized class: sci20.simulator.interfaces.SimulationError

📝 Steps to reproduce

Tried starting a non-blocking Simulation with PyRocky resulting in an error. Since the error is not reported due to the "bug"/missing feature I'm not sure what kind of error at the moment.

💻 Which operating system are you using?

Linux

📀 Which ANSYS version are you using?

25R1

🐍 Which Python version are you using?

3.12

📦 Installed packages

#ansys-rocky-core==0.3.1 !!! Modified to latest working version here + Linux fix !!!
(but I'm 99% sure this is a RyRocky API Translation issue since this error type is not yet implemented)
anyss-tools-path-0.7.1
matplotlib==3.6.3
mpmath==0.0.0
msgpack==1.0.3
numpy==1.26.4
pandas==2.1.4+dfsg
pymeshlab==2023.12.post1
Pyro5==5.15
redis==5.1.1
trimesh==4.4.9
zstd==1.5.5.1

I have additional information regarding the error. Besides the class name also the data pyro5 received:

Exception has occurred: SerializeError
Unsupported serialized class: sci20.simulator.interfaces.SimulationError. Data: {'class': 'sci20.simulator.interfaces.SimulationError', 'exception': True, 'args': ('The simulator is not in an accepted state. Expected it to NOT be in any of "(<SimulationStatus.Executing: 3>,)", but it is on "SimulationStatus.Executing".',), 'attributes': {'_pyroTraceback': ['Traceback (most recent call last):\n', ' File "/home/staff/4fbuuser/ansysdev/bin/Miniforge3/envs/_rocky-5ce45e7/lib/python3.10/site-packages/Pyro5/server.py", line 471, in handleRequest\n', ' File "/home/staff/4fbuuser/ansysdev/agent/_work/3/s/rocky/rocky30/source/python/rocky30/plugins/pyrocky/api_pyro_adapter.py", line 55, in SendToApiElement\n', ' File "/home/staff/4fbuuser/ansysdev/agent/_work/3/s/rocky/rocky30/source/python/rocky30/plugins/pyrocky/api_pyro_adapter.py", line 97, in _SendToObject\n', ' File "/home/staff/4fbuuser/ansysdev/agent/_work/3/s/rocky/rocky30/source/python/rocky30/plugins/api/ra_study.py", line 1353, in StartSimulation\n', ' File "/home/staff/4fbuuser/ansysdev/agent/_work/3/s/rocky/rocky30/source/python/rocky30/plugins/api/ra_study.py", line 1161, in _QuerySimulationService\n', ' File "/home/staff/4fbuuser/ansysdev/agent/_work/3/s/rocky/rocky30/source/python/rocky30/plugins/simulation/rocky_simulation.py", line 783, in StartSimulation\n', ' File "/home/staff/4fbuuser/ansysdev/agent/_work/3/s/rocky/rocky30/source/python/rocky30/plugins/simulation/rocky_simulation.py", line 696, in _SetupSimulator\n', ' File "/home/staff/4fbuuser/ansysdev/agent/_work/3/s/rocky/rocky30/source/python/rocky30/plugins/simulation/rocky_simulation.py", line 1135, in PrepareSimulationAndGetSummary\n', ' File "/home/staff/4fbuuser/ansysdev/agent/_work/3/s/rocky/rocky30/source/python/rocky30/plugins/simulation/rocky_simulation.py", line 1078, in _ConfigureSimulator\n', ' File "/home/staff/4fbuuser/ansysdev/agent/_work/3/s/etk/sci20/source/python/sci20/simulator/base_simulator.py", line 129, in SetInput\n', ' File "/home/staff/4fbuuser/ansysdev/agent/_work/3/s/etk/sci20/source/python/sci20/simulator/base_simulator.py", line 111, in __AssertStatusIsNot\n', 'sci20.simulator.interfaces.SimulationError: The simulator is not in an accepted state. Expected it to NOT be in any of "(<SimulationStatus.Executing: 3>,)", but it is on "SimulationStatus.Executing".\n']}}

---

Regarding actions to get to this specific Simulation Error:
class Rocky:
self.rocky_inst = pyrocky.launch_rocky()
self.project = self.rocky_inst.api.OpenProject(file1)
self.study = self.project.GetStudy()
self.study.StartSimulation(non_blocking=True)
self.project.CloseProject()
self.project = self.rocky_inst.api.OpenProject(file2)
self.study = self.project.GetStudy()
self.study.StartSimulation(non_blocking=True)
--> ERROR

I assumed you get another simulator for a new study but it seems like pyrocky does not support more than one at a time (Before non-blocking it was that limited). Did I do it the wrong way?

Thanks and Greetings
Florian

@Adrian-B-Moreira:
Hello, if possible can you please have a short look or advise we if it's impossible to run more than one simulation at the same time within a single PyRocky instance?
Thanks!

Greetings
Florian

[Adrian-B-Moreira]

Hi @florianreinle , do you mean running multiple simulations using the same port, or are you asking about running multiple separate Rocky instances (with different ports)?

Hi @Adrian-B-Moreira, I mean multiple simulations running the same port.
Since you now can run the simulation non blocking staring a second one (if still solver+hardware available) would make sense.
[Thinking in "UI Style" you close the program but the started simulation still continues (in real UI mode not today since the is the scheduler), start the program and make a new one]

Using more than one port and two Rocky-Instances would be an option as well but atm we have a default port and i would assume adding another port would require a change at Rocky side not PyRocky only.
[Thinking in "UI Style" you start the Rocky two times, start the simulation at both but keep the UI open]

[Adrian-B-Moreira]

@florianreinle I think that using the same port you'll not be able to run multiple simulations, since there's only one Rocky/Freeflow instance running in the background (that could handle only one simulation at a time).

[igortg]

Hi @florianreinle,

As @Adrian-B-Moreira said, it's not possible to run more than one simulation within the same Ansys Rocky instance. The purpose of non-blocking StartSimulation is that you can access partial simulation data (curves) before the simulation finishes.

You may try to run multiple Ansys Rocky instances (if your licenses allows to) in the same script:

rocky1 = pyrocky.launch_rocky(..., server_port=19061)
rocky2 = pyrocky.launch_rocky(..., server_port=19062)

I didn't test that, but it should work.

Hi @igortg

Thank you very much for this idea!
Actually this was the solution what i was looking for, saves me from needing to finalize my more complicated namespace workaround.
Not all problems i have with this are solved at the moment since proxy ownership (and transferring) gets much more complicated this way.
What i can say is parallel pyrocky simulations are definitely working this way.

But this ownership issue is also one point why i still need to start the simulation with "non_blocking".
And due to that we have the exception problem remaining:
I don't get any issues reported that occur during the simulation unlike in the proper rocky scheduler for example. I don't think this is due to pyro5, there might be another reason behind. Or is this error message i was able to get with a code adjustment only possible in the start up phase? And non blocking is never able to report simulation errors during simulation by design?

Greetings
Florian

[femarchi]

Closed by #223