[Darshan-users] Using darshan to characterize PHDF characteristics

[Mahzad Khoshlessan]

Hello all,

I have a python code and I am using parallel hdf5 through h5py. 
I compiled darshan with the following options:
 ./configure --prefix=/home/mkhoshle/installs/darshan --with-mem- align=8 --with-log-path=/oasis/scratch/comet/mkhoshle/temp_project/darshan-logs --with-jobid-env=PBS_JOBID --enable-HDF5-post-1.10 CC=mpicc 

I set the the environment variable LD_PRELOAD as:
export LD_PRELOAD=/home/mkhoshle/installs/darshan/lib/libdarshan.so:/home/mkhoshle/hdf5/v1.10.1/lib/libhdf5.so

And I am using Darshan 3.1.5.
I can run my code with no error but when I run darshan-summary-per-file.sh I get the following error and I get no output.

darshan-summary-per-file.sh /oasis/scratch/comet/mkhoshle/temp_project/darshan-logs/2019/1/15/mkhoshle_python_id10574_1-15-58292-3565400970146387705_1.darshan /oasis/scratch/comet/mkhoshle/temp_project/darshan/58292

Status: Generating summary for file 1 of 64: __init__.py
=======================================================
/home/mkhoshle/installs/darshan/bin/darshan-summary-per-file.sh: line 63: 27764 Segmentation fault darshan-convert --file $hash $1 $2/$file.darshan
=======================================================
darshan-summary-per-file.sh done. Results can be found in /oasis/scratch/comet/mkhoshle/temp_project/darshan/58292/*.pdf.

Am I doing anything wrong? 

Thanks,
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/darshan-users/attachments/20190115/ae210db6/attachment.html>