Error of MPCO recorder in linux

[amb_1]

Dear STKO team,
When I used linux supercomputer system to calculate, the command model ran normally when I deleted mpco record. When I joined mpco, I encountered a new problem, I spent a week, but finally failed. Could you give me some advice?
The links to the opensees and methods I compiled are as follows:
https://github.com/MassimoPetracca/OpenSees
https://opensees.github.io/OpenSeesDocu ... build.html
Best wishes:

[STKO Team]

The STKO_VAR_process_id is automatically defined by STKO in the main.tcl file.
If you just copy-pasted that piece of code inside a file which don't declare that variable you have that error.

Just put this lines at the very beginning of your tcl file:

Code: Select all

# The current process id (from 0 to NP-1)
set STKO_VAR_process_id [getPID]

[amb_1]

Dear STKO Team:
I changed it according to the suggestion, but it still can't run. I used STKO to generate the code, but I don't know where the problem is. Can it run on your liux system (sbatch sub.sh), I have seen Massimo Petracca .PHD running STKO code on designsafe in linux, can you help me to see what is the problem? I have attached the code.
https://www.designsafe-ci.org/rw/worksp ... Simulation
Best wishes:

[STKO Team]

Did you use CMake to compile it?

hdf5.png (20.89 KiB) Viewed 3993 times

It's because you compiled OpenSees with CMake without the MPCORecorder.
To compile OpenSees with the MPCORecorder (and also with the DRM class), you need to install the HDF5 library (version 1.12.0) in a location that CMake can find.

[amb_1]

Dear STKO Team:
When compiling the program code, I used Cmake, is there any other reason? I asked my colleague to follow the tutorial to compile, but it's the same problem, when I delete the mpco record, the program runs normally, but when I join, I can't record the result normally. Can you give me some advice?
https://github.com/MassimoPetracca/OpenSees
https://opensees.github.io/OpenSeesDocu ... build.html
Best wishes:

[STKO Team]

If HDF5 1.12.0 is not installed in a directory that can be found automatically by CMake, you can help CMake find it by specifying the HDF5_ROOT variable as the root of your HDF5 installation.

For example in Windows it is:

hdf5.png (20.89 KiB) Viewed 3979 times

So I would use this:

Code: Select all

cmake .. -DHDF5_ROOT="C:/Program Files/HDF_Group/HDF5/1.12.0"

[amb_1]

Dear STKO team,
As for HDF5, we still have problems after installing HDF5, cmake found HDF5, we studied the source code. In the source code, it defined HDF5 as if def _HDF5, and then executed mpco, but we checked the hdf5 compilation file under linux. There is no definition of marco _HDF5 in it. Then we check the official website file of hdf5 and find that it exists in the shared library of hdf5.lib in windows version. Now the main problem is that we can't find the definition of _HDF5. We tried 1.12.0 and the latest 1.14.0 hdf5, but we couldn't find the definition.Forum You provided the definition of HDF5 for windows. Can you demonstrate the definition of HDF5 for linux?,For example in Linux

[STKO Team]

As for HDF5, we still have problems after installing HDF5, cmake found HDF5, we studied the source code. In the source code, it defined HDF5 as if def _HDF5, and then executed mpco, but we checked the hdf5 compilation file under linux. There is no definition of marco _HDF5 in it

The preprocessor macro _HDF5 is not in HDF5 itself, it is a macro required by OpenSees, so you should defined it in Cmakelist.

You are right, the CMakeList.txt contains an error.
It only defined the _H5DRM macro (used for the DRM object).
It should also define the _HDF5 macro

_hdf5.png (25.28 KiB) Viewed 3882 times

Try this and let us know.

[amb_1]

Dear STKO team,
When using Linux system to Solve by OpenSeesMP, the model ran normally, but when using the site stress balance ，plastic analysis, I encountered a new problem. When the material changed from elastic to plastic after gravity balance, the linux system could not solve, the program is interrupted, but Widows ran normally . have downloaded the folder versions of Widows and Linux last time, can you help me to see it?
Best wishes:
Rujiang Pan

[STKO Team]

1. Did you manage to compile using the _HDF5 macro?

2. the linux system could not solve, the program is interrupted

   Does it crash? or is it just hanging? or not converging?

3. have downloaded the folder versions of Widows and Linux last time, can you help me to see it?

   Are you sure the Windows and Linux version are the same? I mean, you compiled them from the same source code?