The weather model is failing with mpp_domains_define.inc: At least one pe in pelist is not used by any tile in the mosaic
error message
#362
Labels
question
Further information is requested
Is your question related to a problem? Please describe.
While updating the Short-Range Weather Application to stop using the deprecated
atmos_nthreads
and useATM_omp_num_threads
, about half of the comprehensive workflow end-to-end (WE2E) tests are now failing with the following error message:FATAL from PE *: mpp_domains_define.inc: At least one pe in pelist is not used by any tile in the mosaic
This is confusing because there were no issues while using
atmos_nthreads
, there appears to be different behaviors depending on the machine that the updated code is run on, and theufs.configure
,model_configure
, andinput.nml
files look correct.Since there appears to be different behavior depending on the machine (on Hercules, all of the six fundamental tests pass, but one of the tests fails on Hera with the above message), does this suggest that there's an off-by-one or similar edge case based on node size?
Any clarification on what this error message represents and the best way to begin debugging would be greatly appreciated.
Describe what you have tried
From information gleaned off of Google, it looks like this error message occurs when there are issues with either
layout
orio_layout
in theinput.nml
file. To that end, I've made sure that thelayout
entry is properly using the number of MPI tasks to sue in the two horizontal directions (x and y) of the regional grid. These are being properly set. Additionally,io_layout
is being set automatically to1, 1
, which is expected for the SRW App.Any clarification on the error message or suggestions on other things to try and correct this behavior would be greatly appreciated.
Thank you very much for your time.
The text was updated successfully, but these errors were encountered: