I am checking it, before that, are you using a parallel build of the HDF5 library? Thanks. Hong-Jun ----- Original Message -----

From: "Tiantian Xu" <[email protected]> To: "Hong-Jun Kim" <[email protected]> Sent: Thursday, October 13, 2011 12:11:09 PM Subject: Re: moab-announce Digest, Vol 5, Issue 2 Hi Hong-Jun,

Yes, that machine is 64bit, however the Fedora installed is a 32-bit version, I'm not sure if this is the problem... However I have decided not to dig into the problem on that machine any more... I had reinstalled hdf5 and netcdf on a different machine, also 64 bit, with 64-bit Fedora. Moab "configure", "make", "make check" and "make install" were all successful. But the same error in parallel read occurred:

[tiantian@vis-pc2 bin]$ mpiexec -np 2 ./mbconvert -O PARALLEL=READ_PART -O PARTITION=MATERIAL_SET -O PARALLEL_RESOLVE_SHARED_ENTS -O PARALLEL_GHOSTS=3.0.1 -o PARALLEL=WRITE_PART ./64bricks_512hex.h5m dummy.h5m Read "./64bricks_512hex.h5m"

===================================================================================== = BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES = EXIT CODE: 139 = CLEANING UP REMAINING PROCESSES = YOU CAN IGNORE THE BELOW CLEANUP MESSAGES ===================================================================================== APPLICATION TERMINATED WITH THE EXIT STRING: Segmentation fault (signal 11)

I'm very confused because now the "make check" works, all tests are passed, same error still occurs. So it is very likely that there are some subtle issues... The config and make log from the installed lib directory are attached.

Thank you so much for the help.

Tiantian

On Thu, Oct 13, 2011 at 11:33 AM, Hong-Jun Kim < [email protected] > wrote:

Actually, I can not replicate the problem in my machine. Can you check if your machine is 64bit with "uname -m" command?

Could anyone have any idea about the question below? Can "netcdf-4" disabled netcdf cause any segmentation error?

...

Also, before I built MOAB, when I was trying to build netcdf, configure complained it could not find the z library, so I turned off "netcdf-4" feature, in other words, the netcdf library was not built on top of hdf5/zlib. Could this be the problem?

Thanks.

Hong-Jun

...

Hi Hong-Jun,

I still have the same problems after chaning "LFLAGS=-static". "config.log", "mbcn_test.log" and "test-suite.log" are all attached. But I'm not sure what you mean by "error signs".. Also, before I built MOAB, when I was trying to build netcdf, configure complained it could not find the z library, so I turned off "netcdf-4" feature, in other words, the netcdf library was not built on top of hdf5/zlib. Could this be the problem? I'm still trying to fix the "can't find z lib" error with netcdf.

Thanks for your help!

Tiantian

On Wed, Oct 12, 2011 at 1:19 PM, Hong-Jun Kim < [email protected]

...

wrote:

If you still have the same problems with Tim's suggestion as below, could you please send me your "config.log" and error signs for me to replicate it?

In addition, can you use " [email protected] " for this kind of question? "moab-announce" is for announcing new things and upgrades and looks it has some latencies to get mail.(I just got Tim's message which looks sent yesterday)

Thanks.

----------------------------- Hong-Jun Kim Post-doc researcher MCS, Argonne National Laboratory 9700 S. Cass Ave. B240/R2147 Argonne, IL 60439 630-252-4791 [email protected] -----------------------------

...

Hong-Jun, Can you check this out? Tiantian, we haven't seen this problem anywhere. The output from your configure looks like you're on a 64bit machine, but can you verify that? The only other thing that looks strange in your configure is the LFLAGS=-static; you should be able to get the same thing, in a more standard way, using --disable-shared --enable-static. Could you try that? Thanks.

- tim

On 10/11/2011 12:58 PM, Tiantian Xu wrote:

...

Hi,

I was trying to build moab but met with the "mbcn_test fail" during make check. The commands I used to build the library were:

./configure LFLAGS=-static --with-mpi=/home/tiantian/ITL/mpich2-1.4.1p1/gcc --with-hdf5=/home/tiantian/hdf5 --with-netcdf=/home/tiantian/ITL/netcdf-4.1.3/gcc --with-zlib=/home/tiantian/ITL/zlib-1.2.5/gcc --with-szip=/home/tiantian/ITL/szip-2.1/gcc --prefix=/home/tiantian/ITL/moab 2>&1 | tee configure.log make 2>&1 | make.log make check 2>&1 | tee make.check.log

Looks like the mbcn_test is about query on entities, I don't know why this went wrong. Later I ignored the make check errors and did "make install". I tried to perform parallel read/write and found mbconvert went wrong too:

$mpiexec -np 2 ./mbconvert -O PARALLEL=READ_PART -O PARTITION=MATERIAL_SET -O PARALLEL_RESOLVE_SHARED_ENTS -O PARALLEL_GHOSTS=3.0.1 -o PARALLEL=WRITE_PART ./64bricks_512hex.h5m dummy.h5m Read "./64bricks_512hex.h5m"

===================================================================================== = BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES = EXIT CODE: 139 = CLEANING UP REMAINING PROCESSES = YOU CAN IGNORE THE BELOW CLEANUP MESSAGES ===================================================================================== APPLICATION TERMINATED WITH THE EXIT STRING: Segmentation fault (signal 11)

I suspect some fundamental functions in moab were not properly made? I had attached the log files. Please take a look.

Thanks a lot for your help!

Best, Tiantian

-- ================================================================ "You will keep in perfect peace him whose mind is steadfast, because he trusts in you." Isaiah 26:3

Tim Tautges Argonne National Laboratory ( [email protected] ) (telecommuting from UW-Madison)

...

...

phone: (608) 263-8485 1500 Engineering Dr. fax: (608) 263-4499 Madison, WI 53706

------------------------------

_______________________________________________ moab-announce mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/moab-announce

End of moab-announce Digest, Vol 5, Issue 2 *******************************************