Re: [netcdf-hdf] errors building netcdf4 for parallel I/O

NOTE: The netcdf-hdf mailing list is no longer active. The list archives are made available for historical reasons.

yes, HDF5 doesn't have H5Pget_fapl_mpipo. Probably a typo.


At 2:46 PM -0500 9/12/07, Robert Latham wrote:
On Wed, Sep 12, 2007 at 01:20:13PM -0600, Ed Hartnett wrote:
 You must use the --enable-parallel option to configure.

Hi Ed, this didn't seem to make any difference.  config.h has this
alarming section:

/* if true, parallel netCDF-4 is in use */
#define USE_PARALLEL 1

/* if true, compile in parallel netCDF-4 based on MPI/IO */
/* #undef USE_PARALLEL_MPIO */

/* if true, compile in parallel netCDF-4 based on MPI/POSIX */
#define USE_PARALLEL_POSIX 1

When I take a closer look at config.log, the check for MPI-IO (POSIX)
succedds but the check for "real" MPI-IO fails:

configure:29033: checking for H5Pget_fapl_mpiposix in -lhdf5
configure:29068: /home/robl/work/soft/hdf5-1.8.0-beta3/bin/h5pcc -o conftest -g -O2 -I/home/robl/work/soft/hdf5-1.8.0-beta3/include -L/home/robl/work/soft/hdf5-1.8.0-beta3/lib conftest.c -lhdf5 -lhdf5_hl -lhdf5 -lm -lz -lhdf5 >&5
configure:29074: $? = 0
configure:29092: result: yes
configure:29107: checking for H5Pget_fapl_mpipo in -lhdf5
configure:29142: /home/robl/work/soft/hdf5-1.8.0-beta3/bin/h5pcc -o conftest -g -O2 -I/home/robl/work/soft/hdf5-1.8.0-beta3/include -L/home/robl/work/soft/hdf5-1.8.0-beta3/lib conftest.c -lhdf5 -lhdf5_hl -lhdf5 -lm -lz -lhdf5 >&5
conftest.o: In function `main':
/home/robl/work/netcdf-4.0-beta1/build/conftest.c:67: undefined reference to `H5Pget_fapl_mpipo'
collect2: ld returned 1 exit status


On Wed, Sep 12, 2007 at 02:31:23PM -0500, Elena Pourmal wrote:
 Did HDF5 tests pass? There should be no difference between building
 NetCDF-4 and HDF5 tests.....

I can build the HDF5 tests no problem, but the HDF5 tests aren't
trying to use the symbol "H5Pget_fapl_mpipo".  Instead, they use
"H5Pget_fapl_mpiposix"

 Also you may try to remove -O2 flag. The error you are getting,
 doesn't make any sense (at least to me :-) )

The preprocessor suggests that something somewhere is defining
MPI_Comm and MPI_Info to be ints before including mpi.h


Thanks for all the help.  I bet we're pretty close!

==rob

--
Rob Latham
Mathematics and Computer Science Division    A215 0178 EA2D B059 8CDF
Argonne National Lab, IL USA                 B29D F333 664A 4280 315B


--

------------------------------------------------------------
Elena Pourmal
The HDF Group
1901 So First ST.
Suite C-2
Champaign, IL 61820

epourmal@xxxxxxxxxxxx
(217)333-0238 (office)
(217)333-9049 (fax)
------------------------------------------------------------


  • 2007 messages navigation, sorted by:
    1. Thread
    2. Subject
    3. Author
    4. Date
    5. ↑ Table Of Contents
  • Search the netcdf-hdf archives: