Known Issues

XIOS (obelix)

10/11/2016 Some people have found an issue when compiling XIOS in obelix. The error message is normally like this:

Error message:

mpif90 -nofor-main -o xios_server.exe /home/users/yhuang/modipsl/modeles/XIOS/obj/server_main.o -L/home/users/yhuang/modipsl/modeles/XIOS/lib -l__fcm__xios_server -L/usr/local/install/netcdf-4.3.2p/lib -L/usr/local/install/hdf5-1.8.9p/lib -lnetcdf -lnetcdff -lhdf5_hl -lhdf5 -lhdf5 -lz -lcurl /usr/local/install/openmpi-1.6.5/lib
-lstdc++

ipo: warning #11010: file format not recognized for /usr/local/install/openmpi-1.6.5/lib

/usr/local/install/openmpi-1.6.5/lib: file not recognized: Is a directory

fcm_internal load failed (256)

gmake: *** [xios_server.exe] Error 1

gmake -f /home/users/yhuang/modipsl/modeles/XIOS/Makefile -j 1 all failed (2) at /home/users/yhuang/modipsl/modeles/XIOS/tools/FCM/bin/../lib/Fcm/Build.pm line 597

Fix:

[ajornet@obelix2 XIOS]$ svn diff
Index: make_xios
===================================================================
--- make_xios    (revision 858)
+++ make_xios    (working copy)
@@ -157,7 +157,7 @@
XIOS_CINCDIR="$NETCDF_INCDIR $HDF5_INCDIR $MPI_INCDIR"
XIOS_FINCDIR="$NETCDF_INCDIR $XIOS_FINCDIR $MPI_INCDIR"
 
-XIOS_LIB="$XIOS_LIB $NETCDF_LIBDIR $HDF5_LIBDIR $MPI_LIBDIR $NETCDF_LIB $HDF5_LIB $MPI_LIB"
+XIOS_LIB="$XIOS_LIB $NETCDF_LIBDIR $HDF5_LIBDIR $MPI_LIBDIR $NETCDF_LIB $HDF5_LIB -L$MPI_LIB -lnetcdff"
 
rm -f config.fcm
echo "%COMPIL_CFLAGS $COMPIL_CFLAGS" >> config.fcm
Index: arch/arch-ifort_LSCE.path
===================================================================
--- arch/arch-ifort_LSCE.path    (revision 858)
+++ arch/arch-ifort_LSCE.path    (working copy)
@@ -1,6 +1,6 @@
-NETCDF_INCDIR="-I /usr/local/install/netcdf-4.3.2p/include"
+NETCDF_INCDIR="-I/usr/local/install/netcdf-4.3.2p/include"
NETCDF_LIBDIR="-L/usr/local/install/netcdf-4.3.2p/lib"
NETCDF_LIB="-lnetcdf"
-HDF5_INCDIR="-I /usr/local/install/hdf5-1.8.9p/include"
+HDF5_INCDIR="-I/usr/local/install/hdf5-1.8.9p/include" 
HDF5_LIBDIR="-L/usr/local/install/hdf5-1.8.9p/lib"
HDF5_LIB="-lhdf5_hl -lhdf5 -lhdf5 -lz -lcurl"

Added: 10/11/2016

ADA

(21/09/16) ADA IBM MPI library has an unidentified issue with forcesoil. It is necessary to move to Intel MPI library (libIGCM ins_job scripts allows it). When a simulation is launched with more than 32 procs, the upper 32 range must wait for the next MPI call. The lower 32 keep running as usual. In the end, MPI calls are desynchronised. A very long run might create a MPI timeout. So the simulation crashes. In order to make sure it's not Forcesoil fault, it was tested on Curie HPC and ADA Intel MPI. No issues shown.

When debugging in ADA, it showed a weird behaviour. It was not possible to identify the source of this problem.

Routing

This issues are also found in the trunk:

• Resolution at 2 degree does not properly work. Some Nan values might be found in the outputs.
• Non global simulation must not "break" river basins. Be careful when you select EAST_BOUND, WEST_BOUND, ... Otherwise an exception is thrown about -1 index.