Hi 
I am setting up an ACCESS-AM2 model run, and I have gotten an error in the BUILD –> fcm_make_um step. I need to compile from source, but I have gotten this error:
Having a closer look within the suite (u-dn941):
I’m not sure how to fix this, so if anyone has any ideas they would be greatly appreciated !
(sorry for a late Friday afternoon post)
Cheers,
Alanah
Hi Alanah.
This looks like a bash error, i.e. the environment variable OPENMP doesn’t have a value assigned to it at compile time.
In the build script for the rAM3 task fcm_make_um, the value of openmp in app/um/rose-app.conf is hardwired to true.
openmp=true
If you set that in your rose-suite.conf, it should compile.
You’ll then have to figure out why the environment variable that should be assigned a value in your suite is empty.
Thanks Paul! That fixed the error and its compiled!
So it successfully compiled! but now is failing in the atmos step - which I think is still related to openmp?
This is the first error in the job.err file:
With some more detail below:
[0] exceptions: An non-exception application exit occured.[0] exceptions: whilst in a serial region[0] exceptions: Task had pid=109914 on host gadi-cpu-clx-2955.gadi.nci.org.au[0] exceptions: Program is “/home/563/ac9768/cylc-run/u-dn941/share/fcm_make_um/build-atmos/bin/um-atmos.exe”Warning in umPrintMgr: umPrintExceptionHandler : Handler Invokedgc_abort (Processor 0): Job aborted from ereport.
MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLDwith errorcode 9.
NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes.You may or may not see output from other processes, depending onexactly when Open MPI kills them.
[gadi-cpu-clx-2955.gadi.nci.org.au:109801] PMIX ERROR: UNREACHABLE in file /jobfs/35249569.gadi-pbs/0/openmpi/4.0.2/source/openmpi-4.0.2/opal/mca/pmix/pmix3x/pmix/src/server/pmix_server.c at line 2147
Do you think this could be related to setting openmp=true ?
I don’t think so. You have compiled with OMP support but it doesn’t mean you are using it at runtime.
In the rAM3 suite you have to activate it by setting
UM_ATM_OMP=2
in site/nci-gadi/suite-adds.rc
MPI_ABORT is a generic message whenever your MPI process fails/exits.
The error suggests you have incorrect values in the RUN_Nudging nameless section of the UM configuration file.
i.e. this section in /app/um/rose-suite.conf
[namelist:run_nudging]
l_nudging=.false.
!!ndg_analysis_source=
!!ndg_datapath=''
!!ndg_hours_perdata=
!!ndg_lev_bottom=
!!ndg_lev_top=0
!!ndg_on_lev_bottom=0
!!ndg_on_lev_top=0
!!ndg_relax_tvalue=0
!!ndg_relax_uvalue=0
!!ndg_relax_vvalue=0
!!ndg_strat_fac=1.0
Are you trying to run with nudging or not?
Ok, in that case it probably is an issue with RUN_Nudging.
I am trying to run with nudging, but the suite I am using hasn’t been setup to work with ERA5 - although it is possible to run the nudging with ERA5 (done by Fiddes et al 2024); I’ll double check that we’ve set it up properly! We set l_nudging=.true., and ndg_analysis_source=4 (which isn’t defined), so that could be the problem.
Thanks Paul!
Had a typo in the ndg_datapath ! So this error… is fixed