./init_atmosphere_model running problem

[yeamin47]

Hope you are well. I am facing some problem regarding MPAS installation. I have successfully compiled the MPAS. But when I get into step 1.3 Static, terrestrial field processing, I could not understand what to do. As the main instructions for the supercomputer, I have used my common senses and completed the step in there. But when I run the ./init_atmosphere in the terminal I am seeing the following problem.

CRITICAL ERROR: Could not open input file 'x1.10242.grid.nc' to read mesh fields

https://imgur.com/9cgvmuW

https://imgur.com/KemOneY

 

[gdicker]

Hello! To help me check this could you provide a few more things from the run with this error:

1. The streams.init_atmosphere
2. The namelist.init_atmosphere
3. A listing of the files where you ran this attempt? The output of ls or tree will work.

Do you have the 'x1.10242.grid.nc' file in the same directory you ran mpiexec ... ./init_atmophere? What grid file were you intending to use as input to the init_atmosphere core?

(Pictures of your terminal are not very useful for the forum. Copying and pasting text into your post or attachments like your 'Error Log.txt' are preferred.)

 

[yeamin47]

Sad news is that we have deleted the VMware image of that fearing it corrupted the library. From that time period we have tried to compile MPAS v7.3 and failed at library installation. After that, we back to compile V8.2.2. In that case we used built in miniconda library and failed again by some error which is very weird problem. Again, deleted VMware and go back to building library. We successfully built the library but we face problem when compiling MPAS v8.2.1. This time we could not pass compiling let alone reproducing the nc file problem. At this point we are frustrated. Now, we again trying to install Mpas v7.3 from scratch.

N.B. we have successfully installed WRF v4.6 in another VMware and used for hindcast and real time prediction.

 

[gdicker]

Agreed, I've found installs in container-like environments rather frustrating before. If possible, it's been easier to install libraries from source for MPAS-A to control options and troubleshoot. There could be successful installs of MPAS-A using conda for dependencies but I'm currently unaware (users mostly report problems on the forum).

If you aren't already using something similar, the tarballs and iolib_installation.sh script from Index of /people/duda/files/mpas/sources may be helpful. You probably should update the source versions and can modify the script to suit your purposes.

 

[yeamin47]

If we use duda/mpas/sources which mpas version we should choose. That will compatible with the version of duda library

 

[yeamin47]

We have installed all the library perfectly. Lastly, we are facing problem in installing in PIO. We have tried several version such as 2.5.9 | 2.6.2 | 2.4.4 .

I am giving you the log file that we used for cmake. After it was also installed but if we check for its version or finding its path we can not see it (see the PIO path for better understanding)

 

[gdicker]

Apologies for the delay, generally the newest releases of the libraries are recommended:

• Preferred version and flavor of MPI for your use or system
• zlib/1.3.1
• hdf5/1.14.6 built with MPI support
• pnetcdf/1.14.0
• netcdf/4.9.3
• netcdf-f/4.6.1
• OPTIONAL pio/2.6.3 (download from GitHub, 2.6.3 has some bug fixes that are helpful over 2.6.2)

With MPAS-Model v8.0.0 (and newer) the PIO library is now optional. If the PIO environment variable isn't found when compiling, Simple MPAS I/O Layer (SMIOL) will be used instead. It is bundled with the MPAS-Dev/MPAS-Model code since v8.0.0.

I'm not sure what error you were experiencing with PIO. The only information in the attached 'PIO*.txt' files are your commands. The settings seem reasonable, though I haven't tried installing PIO myself in a while.

 

[yeamin47]

Its ok. We have successfully compiled MPAS v8.2. Soon we will publish details guidelines how to do that. As there are no relevant tutorial for that. It was too hectic. But at the end we learnt from our mistakes. Also, thanks for the solution.