3. Compilation and running on Linux

Unable to compile

2020-08-06T16:17:14Z

Hello everyone,
Yesterday I just downloaded the D3DFM source code
(the latest, with highest number). Then I followed the instructions in
src/README to compile it. When it came to CFLAGS='-O2' CXXFLAGS='-O2'
FFLAGS='-O2' FCFLAGS='-O2' ./configure --prefix=`pwd` --with-netcdf
--with-mpi --with-metis --with-petsc, it failed, the final lines are:
(
gfortran: error: unrecognized command-line option '-showme:link'
gfortran: fatal error: no input files
compilation terminated.
configure: creating ./config.lt
config.lt: creating libtool
checking for the extension of Delft3D plugins... .so
./configure: line 20190: ACX_MKL: command not found
checking if PETSc is wanted... yes
checking for PETSc... no
checking for PETSc... no
configure: error: Package requirements (PETSc >= 3.8) were not met:

Package 'PETSc', required by 'virtual:world', not found

Consider adjusting the PKG_CONFIG_PATH environment variable if you
installed software in a non-standard prefix.

Alternatively, you may set the environment variables PETSc_CFLAGS
and PETSc_LIBS to avoid the need to call pkg-config.
See the pkg-config man page for more details.
)
By
the way I'm using gcc & gfortran 10 (I know D3D ask me to use gcc
<8 but I tried many ways to downgrade, just can't, and I can't change
the OS, much more trouble), may be the "gfortran: error:" is because of
this? Other things are, I installed every prerequisites except that
gfortran version I can't go lower. But the terminal showed that the
program was unable to find mpi & petsc. For petsc you can see
clearly from above, for mpi it was a whole dozens of lines, I put them
in the attachment "err.txt". There were something unusual because I
compiled OpenFOAM succesfully (a software that require mpi too).
I tried again and again these two days,
and now I'm hopeless. Did search on forum, there were some questions
quite similar as mine but I can't use those solution. I really need
Delft3D, it is as a treasure to me but, I can't change the OS, how can I
compile successfully help me please.
Forgot to say, I'm using Fedora 32.

RE: tagged revision 65089 not compiling dfmoutput; (missing directory)

2019-10-24T09:38:27Z

Hi Michael,
FLAP is hosted on the external GitHub repository. Make sure that you're not omitting externals while checking out the main delft3d source-code.
Regards,
Dmitry

tagged revision 65089 not compiling dfmoutput; (missing directory)

2019-10-23T16:05:38Z

downloaded tagged revision 65089 as directed.

followed src/README instructions to compile (README is identical to one found in r62441).

dfmoutput does not build because there is no src/third_party_open/FLAP subdirectory in 65089 and dfmoutput depends on that:

mpif90 -DHAVE_CONFIG_H -I. -I../../.. -I../../../utils_lgpl/io_netcdf/packages/io_netcdf/src -I../../../utils_lgpl/deltares_common/packages/deltares_common/src -I../../../utils_lgpl/deltares_common/packages/deltares_common_mpi/src -I../../../third_party_open/FLAP/Test_Driver/mod -DWITH_DELFTONLINE -I/p/app/COST/netcdf-c/4.3.3.1/intel/include -I/p/app/COST/netcdf-fortran/4.4.2/intel/include -I/p/app/COST/netcdf-c/4.3.3.1/intel/include -O2 -fopenmp -I/p/app/COST/netcdf-fortran/4.4.2/intel/include -fpp -recursive -traceback -c -o dfmoutput_main.o dfmoutput_main.F90dfmoutput_main.F90(44): error #7002: Error in opening the compiled module file. Check INCLUDE paths. [IR_PRECISION]use IR_Precision ! Integers and reals precision definition.----^dfmoutput_main.F90(45): error #7002: Error in opening the compiled module file. Check INCLUDE paths. [DATA_TYPE_COMMAND_LINE_INTERFACE]

many, many errors later:compilation aborted for dfmoutput_main.F90 (code 1)make[3]: *** [dfmoutput_main.o] Error 1

imho, it is odd that a tagged revision does not build because of missing source files.

Segmentation fault when running simulation

2018-11-23T13:06:27Z

Hi,

I'm trying to run a simulation from the validation runs with a pre-compiled Dellft3D FM received from Deltares.

However, I get segmentation fault when executing dimr.

Any idea of why is this happening or I what to do to fix it?

Thanks in advance,

Bruna

Unable to run pre-compiled cli version on Linux cluster

2018-11-14T19:42:56Z

Hello,

I am attempting to install a pre-compiled cli version (delft3d-fm-cli-1.0.3-lnx64.tar.gz) that I obtained from Deltares on a Linux cluster. I have run into the following error:

All of the shared libraries required to run the software can be found except the libmpi_cxx.so.1 library, even after setting the LD_LIBRARY_PATH and PATH variables as suggested in the Installation Manual. Please see attached for the ldd command line output. I am using gcc version 6.3.0 and openmpi version 2.0.2, which only contains libmpi_cxx.so.20.0.1, libmpi_cxx.so.20, libmpi_cxx.so, and libmpi_cxx.la.

I assume this problem arises from using a different version of openmpi than was used to compile the program, but I'm not sure how to fix this without compiling from scratch using the source code.

I appreciate any help.

Michelle

RE: Simulation gets stuck in December

2017-04-18T15:21:30Z

Dear Rusty,

Thank you very much for reporting this bug. It is correct to exchange the two lines as you mentioned. We have also corrected the bug in codes of DFlow FM.

With kind regards,

Jing Zhao
Deltares

Simulation gets stuck in December

2017-04-17T20:44:43Z

We have been running annual simulations roughly August-to-August, and with a new model setup the runs get stuck around Dec 1. I believe this is due to this snippet of unstruc.F90, line 12184 in unstruc.F90, rev50237, part of inctime_split. As far as I can tell, newer versions of unstruc.F90 have the same code here.

1
2	case ('M')
3	imonth = imonth + ti_split
4	if (imonth > 12) then
5	imonth = mod(imonth, 12) ; SWAP
6	iyear = iyear + floor(real(imonth)/12.0) ; SWAP
7	end if

Switching the order of the 'SWAP' lines allows the runs to continue, at least in our limited testing so far. Please let me know if I'm missing something. If this is not the correct fix, then perhaps we can post more details of our simulations for some help on figuring out the source of them getting stuck.

Best,
Rusty