Skip to content

Java Segmentation Fault #10158

Open
Open
@Janekdererste

Description

@Janekdererste

Background information

I have posted this to the mailing list as well, but maybe this is the more appropriate place. I am trying to set up a java project with Open-MPI, but I ran into trouble deploying the application on our HPC

What version of Open MPI are you using? (e.g., v3.0.5, v4.0.2, git branch name and hash, etc.)

4.1.2

Describe how Open MPI was installed (e.g., from a source/distribution tarball, from a git clone, from an operating system distribution package, etc.)

I installed ompi from a tarball, using the following commands:

$ tar xf openmpi-<version>.tar.bz2
$ cd openmpi-<version>
$ ./configure --prefix=<path/to/where/the/installation/should/go> --enable-mpi-java --with-jdk-dir=/path/to/desired/jdk/
$ make -j 8 
$ make install

Please describe the system on which you are running

  • Operating system/version: AlmaLinux release 8.5 (Arctic Sphynx)
  • Computer hardware: AMD EPYC 7302 16-Core Processor, 32 cores, 1007G
  • Network type: Infiniband

Details of the problem

I receive a segmentation faul when our logging library is initialized. The following program leads to such a chrash:

package debug;

import mpi.MPI;
import mpi.MPIException;
import org.apache.logging.log4j.LogManager;

public class LoggerExample {

    public static void main(String[] args) throws MPIException {

        MPI.Init(args);
        var logger = LogManager.getLogger(LoggerExample.class);
        MPI.Finalize();
    }
}

The example can be found here. To run the above example the program can be build with:

mvn clean install

and then be executed with:

mpirun -np 1 java -cp matsim-p-1.0-SNAPSHOT.jar debug.LoggerExample

The core dump of the VM is here core-dump-loggerExample.log

As suggested by Benson on the mailing list, I also tried to run the ping-pong-mpi-tcp project. The core dump of this crash is here stack-trace-ping-pong.log. Both programs crash while conducting String operations.

I have tried different things so far:

  • On my local installation WSL-Ubuntu 20.04, both projects run fine.
  • On another HPC configuration (Host: Intel(R) Xeon(R) Silver 4210 CPU @ 2.20GHz, 40 cores, 376G, CentOS Linux release 7.9.2009 (Core)) the same crash is happending
  • I tried different Java-Versions which didn't work, (Java 11 and 17)
  • I tried to compile the code with mpijavac, This also didn't work. As far as I can see, this command only adds mpi.jar to the classpath, this should be accomplished by the maven set up in the example project.

Any help would be much appreciated!

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions