Issues with METplus Install (possibly with build step?)

[jhicks14]

Hello,

I am trying to install the latest versions of METplus/MET locally. I have been using a shared NOAA install, but I am interested in downloading/installing to my local NOAA machine so that I have more control over updates and such.

If followed the install procedures on the installation guide: https://metplus.readthedocs.io/en/latest/Users_Guide/installation.html, but I am not having any luck running METplus after completing all steps. I've changed the MET_INSTALL_DIR, CONFIG_DIR, and INPUT_BASE in my configuration file (otherwise it's the same config file I used for the original METplus install), but running run_metplus.py begins to run METplus and then freezes. I believe the issue may lie with the build script, but I'm not entirely sure (there seem to be a few errors early in the build run, but they may be harmless?).

I tried re-installing, but still had no luck.

[jprestop]

Hi @jhicks14. I see you are having trouble with the installation of the METplus and MET packages.

In the link to the METplus Installation Guide, I see it leads to the Software Installation section of the MET User's Guide, where it says "Users can find the script and its instructions under on the Downloads page of the MET website."

Can you please confirm that it is the compile_MET_all.sh script from the Downloads page that you are using to install MET?

[jprestop]

HI @jhicks14. A colleague of mine just pointed out some instructions further down on the METplus Installation Guide:
https://metplus.readthedocs.io/en/latest/Users_Guide/installation.html#build-components-and-using-manage-externals

This leads me to believe you were using the build_MET.sh script for installation instead of the compile_MET_all.sh script. MET is intended to be compiled by users from a tarfile, not directly from the repository, which is why our Downloads links directly to the tarfile on GitHub that users should compile. We do development within the repository but run a script to create a release. That script runs some tools to automatically generate some code for utilities that are needed for development but not for compiling the released versions.

To compile MET, please refer to the section titled:
Sample Script For Compiling External Libraries And MET

On the MET downloads page:
https://dtcenter.org/community-code/model-evaluation-tools-met/download

We realize that the presence of the build_MET.sh script in the METplus repository is confusing. It is included as part of our work toward the use of manage_externals, but unfortunately, it isn't working as expected. In fact, that script does just call the compile_MET_all.sh script I've referenced above. Ideally, that would all just work out of the box, but it looks like we'll need to work on refining it some more.

[jhicks14]

Good morning,

Thanks for the recommendations! I think I almost have this fully figured out. The only thing that seems to be missing after using the compile_MET_all.sh script (and passing in a file with the environmental variables) is the /bin directory that contains the MET tools such as plot_data_plane.

Does this indicate that I set my environmental variables incorrectly?

[jprestop]

Hi @jhicks14. It's hard to say why the executables were not built. It would be helpful to look at your config.log file and a make.log file, which should be in your met-10.0.0 directory, and also your configuration file for the compile_MET_all.sh script. Could you please send those to us, either by attaching them here (if it won't let you attach the configuration file for the compile_MET_all.sh script because of file extension, you can just add a .txt to the end of the filename) or by following the "How to Send Us Data" section of the Resources for Troubleshooting post to send them via ftp? Thanks!

EDIT: The user sent data. Unfortunately, this issue was not reproducible and has not been encountered previously. It was suggested that the user contact the NetCDF helpdesk for support. See the comments below for details, however, the user decided to produce a different route.

[jhicks14]

Still working on it! Had to put this on the back-burner for now while I finish up some other projects.

[jprestop]

@jhicks14 That's no problem. Thank you very much for the update.

[jprestop]

Hi @jhicks14. I'm guessing this work may still be on the back-burner but just wanted to check in about it, especially since I can't recreate this problem, haven't seen it before, and am stumped by it. :)

[jhicks14]

Hi @jprestop. Since I was having issues installing MET locally, we just moved a previous install to a directory that is accessible by more users.

For version control, I think we'll get a system administrator to install the newest MET versions when needed. I think we can close this thread for now, and then I'll reach out if we have any more issues with future installs!

Thanks again!

[jprestop]

Thank you, @jhicks14. I have modified the comment above in order to be able to choose an "Answer" for this question. Please let me know if you'd like me to make any changes. Otherwise, I'll go ahead and lock this discussion.