search

E3SM-Project / scream

Fork of E3SM used to develop exascale global atmosphere model written in C++
https://e3sm-project.github.io/scream/
Other
80 stars 57 forks source link

Wrong variable name in shoc standalone input file #979

Closed bartgol closed 3 years ago

bartgol commented 3 years ago

This is causing master failures. I'm not sure how it got past the AT.

See comment at bottom of #973 for details.

bartgol commented 3 years ago

Ok, I went on mappy and manually checked the test logs for PR #974 , which is marked as "passsed". Inside ctest's log, I can see this:

 "shoc_stand_alone_ut_np1_omp1" start time: Apr 05 09:16 MDT 
 Output:
 ----------------------------------------------------------
 ExecSpace name: OpenMP
 ExecSpace initialized: yes 
  avx -AVX2-AVX compiler GCC default FPE mask 0 (NONE)
  #host threads 1
 sizeof(Real) = 8 
 default pack size = 16

      -49                                                                                                                                       PIO ERROR: could not find variable T_mid in file shoc_standalone_ne4_ne4.nc
 --------------------------------------------------------------------------
 MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD
 with errorcode 0.

Clearly, the error is happening. However, I suspect we call MPI_Abort with the wrong error code (0!!). When 0 is then returned to the caller (ctest) it is interpreted as a no-error sign.

bartgol commented 3 years ago

Ok, this line is the problem. The 2nd input of mpi_abort should be the error code, while the 3rd is simply the error code _returned by mpiabort itself.

I will put up a ninja fix PR.