adding new machine

[jjakacki@iopan_gda_pl]

Dear users,i am trying to add new machine into cesm 1.2.0 but when i use cesm_setup script it does not have EXEROOT variable which is defined in env_build.xml file. I spent a lot of time on trying to diagnose the problem but without any progress. I will be glad if somebody will give me advice hot to fix this problem.With best regards,Jaromir p.s. i have added flag -v in the preview_namelist script and it looks like it does not read any variables from env_build.xml file

 

[jedwards]

Is it possible that the file has been corrupted and is no longer recognised as xml?  

 

[jjakacki@iopan_gda_pl]

Thank you for the replay,I have just figured out the problem. in the config_machines.xml are two paths. EXEROOT and RUNDIR. EXEROOT has to have in the end ../bld and i put full path for the RUNDIR. If it is defined by EXEROOT it did not work. I have one simple questions - why do you still make the model more compliated? Maybe you should start to work on better validation than making more cases? I understand that the project have a lot of goals and you are making plan, but it is possible that in the end of the project only developers will be able to work with. I think it is very good idea to make model like this, but also it is much better if the model will have more users.Again thanks for trying to help,with best regards,jaromir

 

[jjakacki@iopan_gda_pl]

hello dears again,i am fighting with porting the model. I am not administrator, so it gives me a lot of pain. After fixing another problem with precompailer, i got an error when trying to compile pio: mpif90  -c -DLINUX -DMCT_INTERFACE -DHAVE_MPI -DFORTRANUNDERSCORE -DNO_R16 -fpp -DLINUX -DSEQ_ -DFORTRANUNDERSCORE -DNO_SHR_VMATH -DLINUX -DC
PRINTEL -DHAVE_SLASHPROC -DSPMD -DHAVE_MPI -DNO_C_SIZEOF -DSYSLINUX -D_NETCDF -I/users/kdm/mjanecki/netcdf4/include -D_NOPNETCDF -DTIMING  -O
2 -fp-model source -convert big_endian -assume byterecl -ftz -traceback  -free  -I.  -I/users/tmp/jakacki/cases/STUB01/bld/pio -I/users/kdm/m
janecki/netcdf4/include -I/users/tmp/jakacki/cases/STUB01/bld/mct/mct -I/users/tmp/jakacki/cases/STUB01/bld/mct/mpeu -I/users/tmp/jakacki/cas
es/STUB01/bld/pio -I/users/tmp/jakacki/cases/STUB01/bld/gptl -I/users/kdm/jakacki/CESM/UCAR/cesm1_2_0/models/utils/pio -I/users/tmp/jakacki/c
ases/STUB01/bld/lib/include -I/users/kdm/mjanecki/netcdf4/include   -I../timing  /users/kdm/jakacki/CESM/UCAR/cesm1_2_0/models/utils/pio/piol
ib_mod.F90
/users/kdm/jakacki/CESM/UCAR/cesm1_2_0/models/utils/pio/piolib_mod.F90(1959): error #6404: This name does not have a type, and must have an e
xplicit type.   [MPI_ROOT]
                iosystem(i)%iomaster = MPI_ROOT
---------------------------------------^
/users/kdm/jakacki/CESM/UCAR/cesm1_2_0/models/utils/pio/piolib_mod.F90(2092): warning #6843: A dummy argument with an explicit INTENT(OUT) de
claration is not given an explicit value.   [IOPROC]
  subroutine pio_recommend_iotasks(comm, ioproc, numiotasks, miniotasks, maxiotasks )
-----------------------------------------^
/users/kdm/jakacki/CESM/UCAR/cesm1_2_0/models/utils/pio/piolib_mod.F90(2092): warning #6843: A dummy argument with an explicit INTENT(OUT) de
claration is not given an explicit value.   [NUMIOTASKS]
  subroutine pio_recommend_iotasks(comm, ioproc, numiotasks, miniotasks, maxiotasks )
-------------------------------------------------^
compilation aborted for /users/kdm/jakacki/CESM/UCAR/cesm1_2_0/models/utils/pio/piolib_mod.F90 (code 1)
gmake[1]: *** [/users/kdm/jakacki/CESM/UCAR/cesm1_2_0/models/utils/pio/piolib_mod.o] Error 1
gmake[1]: Leaving directory `/users/tmp/jakacki/cases/STUB01/bld/pio'
gmake: *** [all] Error 2 I will be glad if somebody will be able to help me,with best regards,jaromir

 

[jjakacki@iopan_gda_pl]

I have found on the list that flag -DNO_MPI2 fixes this problem and it works. But i got another error during compilation of pio: /users/kdm/jakacki/CESM/UCAR/cesm1_2_0/models/utils/pio/pio_support.F90(355): error #6645: The name of the module procedure conflicts with a
name in the encompassing scoping unit.   [MPI_TYPE_CREATE_INDEXED_BLOCK]
  subroutine MPI_TYPE_CREATE_INDEXED_BLOCK(count, blen, disp, oldtype, newtype, ierr)
-------------^
compilation aborted for /users/kdm/jakacki/CESM/UCAR/cesm1_2_0/models/utils/pio/pio_support.F90 (code 1)
gmake[1]: *** [/users/kdm/jakacki/CESM/UCAR/cesm1_2_0/models/utils/pio/pio_support.o] Error 1
gmake[1]: Leaving directory `/users/tmp/jakacki/cases/STUB01/bld/pio'
gmake: *** [all] Error 2 it would be great if somebody will help me.with best regards,jaromir I did some test and flag DNO_MPI2 helps with the problem: iosystem(i)%iomaster = MPI_ROOT but still MPI_TYPE_CREATE_INDEXED_BLOCK exist (looks like circle)with best,j.    

 

[jedwards]

The -DNO_MPI2 flag should not be needed with any reasonably modern MPI version - what version are you using?    Look in mpif.h and see if MPI_ROOT is defined - if it is then check that your mpif90 wrapper script is working correctly.

 

[jjakacki@iopan_gda_pl]

Thank you jedwards,i have passed the compilation process. I change PE (as you suggested to the newest one) and compilation is done. But when i submitted the jobs (dead (X) and stub (S) compsets) the model will not start because it could not find executable file. The error message is: STUB01/run$ more ccsm.log.140122-200141
--------------------------------------------------------------------------
mpirun was unable to launch the specified application as it could not find an executable:

Executable: -o
Node: g472

while attempting to start process rank 0.
-------------------------------------------------------------------------- I think i did something wrong in directiores names. The executable file exist in the bld directory, but it is not going to start. Could somepbody pleast point me the directory structure of the cesm1.2 model (of course only executable, compilation and running part).thank you in advance for any helpregards,jaromir  

 

[jjakacki@iopan_gda_pl]

Finally i got the model working. I tested two compsets: X and S and it works fine.Thank you for your help,with best regards,jaromir