alan_m_rhoades@gmail_com
New Member
Hello,I'm currently trying to generate mapping files for CLM4.5 using the CESM1.2 shared toolsets on Yellowstone. I have successfully done this for CLM4.0 for two seperate SCRIP files that my research team has generated. Our usual order of operations for creating the CLM4.0 mapping files were:---------------------------------------1) cd to ${HOME}/models/lnd/clm/tools/shared/mkmapdata2) vi into mkmapdata.sh and comment out grid name 10x10min_IGBmergeICESatGIS (hangs up clm4_0 map file generation) and other map file names that have been previously generated (to speed up file creation process)3) On Yellowstone enter an interactive mode to execute mkmapdata.sh*EXAMPLE (for one of our team's grid) AR_calif_x4: execgy ./mkmapdata.sh -v --phys clm4_0 --gridfile AR_calif_30_x4.g_scrip.nc --res AR_30_x4 --gridtype global*NOTE: should take a couple of hours to finish - need to generate ~12 mapping files--------------------------------------I tried applying this same process to the CLM4.5 mapping files and generated all of them EXCEPT for one, map_1km-merge-10min_HYDRO1K-merge-nomask_to_AR_30_x4_nomask_aave_da_c150128.nc. I've looked at similar posts throughout the forum and have seen that this is probably due to the fact that there is insufficient memory on the node I'm trying to submit the shell script to.So, I tried creating a bash script to submit to the geyser queue (because I was unsure on how to best use the regridbatch.sh file to incorporate our newly generated SCRIP file). This was the content of the bash script that I generated:--------------------------------------#!/bin/bash## Currently only setup to run on yellowstone/caldera/geyser. Note that# geyser is needed for very high resolution files (e.g., 1 km) because# of its large memory per node, so that is set as the default.# However, for coarser resolutions, you may get better performance on# caldera or yellowstone.## yellowstone specific batch commands:#BSUB -P#BSUB -n 8#BSUB -R "span[ptile=8]"#BSUB -o regrid.%J.out # ouput filename#BSUB -e regrid.%J.err # error filename#BSUB -J regrid # job name#BSUB -W 20:00#BSUB -q geyser # queue ./mkmapdata.sh -v --batch --phys clm4_5 --gridfile AR_calif_30_x4.g_scrip.nc --res AR_30_x4 --gridtype global---------------- However, when I submit this script with the following command line argument... ---------------- [arhoades@yslogin5 mkmapdata]$ bsub -q geyser -W 20:00 -n 1 -P ******** geysersubmit.shJob is submitted to queue .[arhoades@yslogin5 mkmapdata]$ bjobsJOBID USER STAT QUEUE FROM_HOST EXEC_HOST JOB_NAME SUBMIT_TIME89331 arhoade PEND geyser yslogin5-ib *submit.sh Jan 28 15:20[arhoades@yslogin5 mkmapdata]$ bjobsNo unfinished job found----------------------------------- The submission script goes to the geyser queue and is run, but then quickly dies and doesn't generate error or log files. Does anyone have any insights on how best to move forward on this? It seems like all of the mapping files necessary to generate the CLM4.5 surface datasets were created successfully, just need a work around for the memory allocation issue for the aforementioned 1km dataset. All the best, AR