STAR-CCM+ OpenMPI Job Failed due to Out-of-Memory

After the scheduled downtime on May 12, 2026, STAR-CCM+ has been encountering out-of-memory errors when running OpenMPI jobs. A message similar to the following appears repeatedly in the output:

[c0317.ten.osc.edu:1736021] Read -1, expected 133860, errno = 1

We have not yet identified the root cause of the issue; however, we have discovered that this error does not occur in multi-node jobs.

Workarounds

Option 1: Convert to a Multi-Node Job

Users can temporarily resolve this by converting a single-node OpenMPI job into a multi-node job.

Option 2: Switch to Intel MPI

Alternatively, users can switch to Intel MPI by setting the following environment variables before executing the starccm+ command:

export -n I_MPI_HYDRA_BOOTSTRAP_EXEC_EXTRA_ARGS
export I_MPI_HYDRA_BOOTSTRAP=ssh
export FI_PROVIDER=verbs