Error with bin folder #1849
-
|
Replace italics below with details for this issue. Describe the ProblemI am trying to run a test cast with the MODE. I have the all the input files and the directory get an error every time about the location of the .../bin/mode not being found. I don't have this problem with any other path. I tried adding the directories myself but got this error now PermissionError: [Errno 13] Permission denied: '/gpfs/fs1/home/ac.jcorner/NIU/Masters_Thesis/metplus/bin/mode EDIT Expected BehaviorFor the test case to work EnvironmentRun on Argonne's computer Bebop cluster in a Linux environment To Reproducepython run_metplus.py -c /gpfs/fs1/home/ac.jcorner/NIU/Masters_Thesis/metplus/parm/use_cases/met_tool_wrapper/MODE/MODE.conf Relevant DeadlinesThis Thursday I have it due to my advisor |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 30 replies
-
|
Hi @jcorner1, I converted this issue to a discussion because that is where we provide help support. We use the issues to track new development. It sounds like you are having issues running the METplus wrappers. The first error you are receiving seem to indicate that the MET_INSTALL_DIR METplus config variable that controls where to find the MET executables is not set properly. How did you obtain the METplus wrappers? You can run Regardless if you are able to modify the defaults.conf file or not, we recommend creating a User Configuration File to store settings specific to you and the environment you are using. You can set the correct value for MET_INSTALL_DIR in this file, along with other settings like OUTPUT_BASE. You can pass your user config file to run_metplus.py in addition to the use case config file. We recommend listing it last so that the values set in that file take priority:
However, the 2nd error you received seems to indicate that you do not have permissions to run the mode MET executable. If these binaries were created by another user, you may need to reach out to them to open up the permissions. |
Beta Was this translation helpful? Give feedback.
-
|
Hi @jcorner1. Thanks for letting me know. It sounds like your computer does use modules. While you could use the already installed software on that machine, you don't have to. The only thing necessary would be a loaded compiler and the path for that is not necessary. You could run, for example, "module avail intel" or "module avail gnu" to see what compilers are available to load. If you'd like to enable python embedding you can see what version of python are available by running "module avail python". While some paths will be needed for python, I may be able to help you figure out what those are. Feel free to post the compiler error you're receiving here and I'll see if I can help. Regarding a zoom call, I've reached out to my project manager to see if a zoom call would be possible, but am waiting to hear back. I will follow up once I know more. In the meantime, I'm happy to keep assisting in this Discussion. |
Beta Was this translation helpful? Give feedback.
-
|
Good news @jprestop, it would appear I am very close to getting met to successfully run. After some help from the Argonne help desk and the paths and variables look to be set correctly or least on the right track. I ran the |
Beta Was this translation helpful? Give feedback.
-
|
That's great news, @jcorner1. Do you have the met-10.1.2.20220516.tar.gz file in the tar_files directory? That is where the script is expecting to find it. If so, please upload your compile_MET_all.sh script, your install_met_env file you're using to install, and the screen output from your run of compile_MET_all.sh to our ftp server using the instruction in the How to Send Us Data post. Please let me know once you've done that so that I know to grab the files. Thanks! |
Beta Was this translation helpful? Give feedback.
-
|
@jprestop I tried to access the server but wasn't able to do so through putty. I was able to do so through WinSCP but am unable to find where to locate my files in the directories that I see. UPDATE |
Beta Was this translation helpful? Give feedback.
-
|
Hi @jcorner1. Thank you for sending up those files. I received the script and the configuration files. Could you also please send the screen output from your run of compile_MET_all.sh script? Could you also please post the result of running "ls /gpfs/fs1/home/ac.jcorner/NIU/Masters_Thesis/MET_10.1.2/tar_files"? On a separate note, if you end up using the install_met_env.generic file, it looks like the value for MET_PYTHON_LD will need to be changed. It is currently set in the following way: There is a line above that one that is set like this, but is commented out: That is closer to what you'd need but is still missing the "-lpython3.9" and is missing backslashes before the spaces (alternatively you can leave out the backslashes and put the value in quotation marks). This should be what you need: The same is true for the install_met_env.cray file in that the value of MET_PYTHON_LD will need to be changed. |
Beta Was this translation helpful? Give feedback.
-
|
Hi @jprestop Here is the output I get when I run the compile and .generic file. I updated the .generic file to have the path you gave which definitely was helpful and didn't have an error I was having when I ran the commented out path. We seem to be very close. Here is the screen output from running the ls command in the tar_files dir: Furthermore, the The original output I was given when running the However I changed it like I said because I was getting an error when running it. Let me know if there is something that needs to be changed in the MET_PYTHON_CC path as well. Thanks! |
Beta Was this translation helpful? Give feedback.
-
|
Hi @jcorner1. Thanks for sending this information. I'm glad we are getting closer! I see the following in your make.log file: libsmvl is the Intel C Compiler runtime library. Could you please run "module list" to see if Intel is still loaded? If it is could you please try using a clean login/shell so that it is no longer loaded? I see the following in the config.log file: I think if we can get those out of your PATH environment variable that might do the trick. I'm hoping that getting a new login and clean shell will remove those from your PATH. After you get a new shell, could you please run "echo $PATH | grep -i intel" to see if any Intel paths still exist? Thanks! Please let me know how it goes. |
Beta Was this translation helpful? Give feedback.
-
|
Hi @jprestop I ran a module unload intel/17.0.4-74uvhji which moved all the intel modules into a inactive list of modules. After running the "echo $PATH | grep -i intel" command and no paths popped up so I presumed they were removed. However, after running the script they paths seem to be appearing in the .log files. I'll attach the files. A little clarification on what is meant by a clean login/shell might help. Does this mean logging into the server without the intel modules loading? If so, any advice on how to do so? I'm unsure on how to proceed. Thanks! |
Beta Was this translation helpful? Give feedback.
-
|
Hi @jcorner1. Thank you for running the module unload command. I think that did actually work as I don't see any intel paths in your config.log file. I think I have an idea of what is going on here. After we switched to having you use the GNU set of compilers, I do no believe I told you to recompile all of the external libraries needed for MET. So, what I suspect is happening here is there is a linking problem with MET the external libraries. It's actually a good thing that you are seeing this error, because it is possible that you could have had a successful compilation, but then experienced some strange run-time problem. MET and its dependent libraries need to be compiled with the same "family" of compilers or else strange and unexpected things could happen. So, please comment out the following in your install_met_env.generic.txt file: by adding a "#" in front of each line (and MET_GSL and MET_HDF5, if you happen to have set those since the last configuration file you sent me). This will tell the compilation script to reinstall those libraries using the GNU set of compilers. Please let me know how it goes. And, regarding "A little clarification on what is meant by a clean login/shell might help. Does this mean logging into the server without the intel modules loading? If so, any advice on how to do so?" I simply meant to logout and log back in again to get a clean shell, however, it is possible that in your .bashrc or .cshrc (or other) file that modules were being loaded and the "clean" shell wouldn't have helped, but it looks like the "module unload" accomplished what was needed. At least I think so... :) |
Beta Was this translation helpful? Give feedback.
-
|
Hi @jprestop, Great News! We have a successful compilation of met! Onto the installation of METplus now. Thank you so much for your help and patience in compiling met. If I have any questions installing METplus, should I ask them in this thread or in a different discussion post? or start a new one? Thanks! |
Beta Was this translation helpful? Give feedback.
-
|
Hi @jcorner1. That certainly is great news! Thank you for letting us know. I'm glad we were able to work through all of this, resulting in a successful compilation. Thank you for asking about where to ask questions about future work. I’m going to LOCK this conversation to prevent future posts in this thread. Our team has decided that once discussions have been answered, and the answer has been confirmed, we’ll lock them. We want to encourage users to ask new questions in new discussions rather than posting to old ones. Hopefully that’ll make the questions/answers easier for other users to follow. So if/when more issues/questions arise, please feel free to start a new discussion. |
Beta Was this translation helpful? Give feedback.
Hi @jprestop
Here is the output I get when I run the compile and .generic file. I updated the .generic file to have the path you gave which definitely was helpful and didn't have an error I was having when I ran the commented out path. We seem to be very close.