This section provides the troubleshooting information on typical MPI failures with corresponding output messages and behavior when a failure occurs.
If you encounter errors or failures when using the Intel® MPI Library, take the following general troubleshooting steps first:
- Check the Intel® MPI Library System Requirements section and the Known Issues section in the Intel® MPI Library Release Notes.
- Check accessibility of the hosts. Run a simple non-MPI application (for example, the hostname utility) on the problem hosts using mpiexec. For example:
> mpiexec -ppn 1 -n 2 -hosts node01,node02 hostname node01 node02
This may help reveal an environmental problem, or a connectivity problem (such as unreachable hosts).
- Run the MPI application with debug information enabled: set the environment variables I_MPI_DEBUG=6 and/or I_MPI_HYDRA_DEBUG=on. Increase the integer value of debug level to get more information. This action helps narrow down to the problematic component.
- If you have the availability, download and install the latest version of Intel MPI Library from the official product page and check if your problem persists.
- If the problem still persists, you can submit a ticket via the Support page, or ask experts on the community forum.