Update entrypoint for partition config checking.

Updating once more to make sure logic for checking for parallel
processing is correct where needed, and then simplifying subsequent
logic that can rely on results of previous check.

docker/main/ngen/entrypoint.sh

@@ -49,9 +49,15 @@ OUTPUT_DATASET_DIR="${ALL_DATASET_DIR}/output/${OUTPUT_DATASET_NAME}"
 HYDROFABRIC_DATASET_DIR="${ALL_DATASET_DIR}/hydrofabric/${HYDROFABRIC_DATASET_NAME}"
 REALIZATION_CONFIG_DATASET_DIR="${ALL_DATASET_DIR}/config/${REALIZATION_CONFIG_DATASET_NAME}"
 BMI_CONFIG_DATASET_DIR="${ALL_DATASET_DIR}/config/${BMI_CONFIG_DATASET_NAME}"
-# Don't require a partitioning dataset when only using a single node
-if [ ${MPI_NODE_COUNT:?} -gt 1 ]; then
+# Check if parallel processing is in effect and partition dataset is needed by testing node count or 1st node CPU count
+if [ ${MPI_NODE_COUNT:?} -gt 1 ] || [ $(echo "${MPI_HOST_STRING}" | sed 's/,//' | awk -F: '{print $2}') -gt 1 ] 2>/dev/null; then
     PARTITION_DATASET_DIR="${ALL_DATASET_DIR}/config/${PARTITION_DATASET_NAME:?No partition config dataset name for directory}"
+# Note that, if the above test is "false" (in particular, the CPU count check) we should ensure the host string is valid
+# Catch false negative due to invalid CPU count/format by taking complement of whether 1st CPU count is greater than -1
+# Any bogus value will result in the pre-complemented test being "false"
+elif ! [ $(echo "${MPI_HOST_STRING}" | sed 's/,//' | awk -F: '{print $2}') -gt -1 ] 2>/dev/null ; then
+    echo "Error: invalid CPU count parsing for first host of MPI host string '${MPI_HOST_STRING}'" 2>&1
+    exit 1
 fi
 
 RUN_SENTINEL="/home/${MPI_USER}/.run_sentinel"
@@ -168,17 +174,8 @@ exec_serial_ngen_run()
 # Sanity check that the output, hydrofabric, and config datasets are available (i.e., their directories are in place)
 check_for_dataset_dir "${REALIZATION_CONFIG_DATASET_DIR}"
 check_for_dataset_dir "${BMI_CONFIG_DATASET_DIR}"
-# Require a partitioning dataset only when doing parallel, multiprocess ngen job with MPI
-# Most obvious indicator of parallel processing (and simplest to check) is if there are multiple host nodes
-if [ ${MPI_NODE_COUNT:?} -gt 1 ]; then
-    check_for_dataset_dir "${PARTITION_DATASET_DIR:?No partition dataset directory defined}"
-# If using only a single node, then we must check how many CPUs the host has
-elif [ $(echo "${MPI_HOST_STRING}" | sed 's/,//' | awk -F: '{print $2}') -gt 1 ] 2>/dev/null ; then
-    check_for_dataset_dir "${PARTITION_DATASET_DIR:?No partition dataset directory defined}"
-# Also sanity check the host string format and CPU count extraction, ensuring we didn't produce a false negative above
-elif [ $(echo "${MPI_HOST_STRING}" | sed 's/,//' | awk -F: '{print $2}') -ne 1 ] 2>/dev/null ; then
-    echo "Error: failed to extract integer CPU count for first host of MPI host string '${MPI_HOST_STRING}'" 2>&1
-    exit 1
+if [ -n "${PARTITION_DATASET_DIR:-}" ]; then
+    check_for_dataset_dir "${PARTITION_DATASET_DIR}"
 fi
 check_for_dataset_dir "${HYDROFABRIC_DATASET_DIR}"
 check_for_dataset_dir "${OUTPUT_DATASET_DIR}"