fix: upgrade Intel MPI to 2021.14 and fix CI race condition

Upgrades Intel MPI version from 2021.13 to 2021.14.

Adds a 5-second initialization delay in entrypoint.sh to prevent SSH handshake failures (kex_exchange_identification) observed in CI environments.

Fixes Issue #678.

Signed-off-by: Syed-Suhaan <[email protected]>

Author: Syed-Suhaan

build/base/entrypoint.sh

@@ -31,6 +31,8 @@ if [ "$K_MPI_JOB_ROLE" == "launcher" ]; then
   do
     resolve_host "$host"
   done
+  # Add a delay to allow worker's sshd to be fully ready in slow CI environments
+  sleep 5
 fi
 
 exec "$@"

build/base/intel-builder.Dockerfile

@@ -16,9 +16,9 @@ RUN apt update \
     && echo "deb [signed-by=/usr/share/keyrings/oneapi-archive-keyring.gpg trusted=yes] https://apt.repos.intel.com/oneapi all main" | tee /etc/apt/sources.list.d/oneAPI.list \
     && apt update \
     && apt install -y --no-install-recommends \
-        libstdc++-12-dev binutils procps clang \
-        intel-oneapi-compiler-dpcpp-cpp \
-        intel-oneapi-mpi-devel-2021.13 \
+    libstdc++-12-dev binutils procps clang \
+    intel-oneapi-compiler-dpcpp-cpp \
+    intel-oneapi-mpi-devel-2021.14 \
     && apt remove -y gnupg2 ca-certificates apt-transport-https \
     && apt autoremove -y \
     && rm -rf /var/lib/apt/lists/*

build/base/intel.Dockerfile

@@ -18,8 +18,8 @@ RUN apt update \
     && echo "deb [signed-by=/usr/share/keyrings/oneapi-archive-keyring.gpg trusted=yes] https://apt.repos.intel.com/oneapi all main" | tee /etc/apt/sources.list.d/oneAPI.list \
     && apt update \
     && apt install -y --no-install-recommends \
-        dnsutils \
-        intel-oneapi-mpi-2021.13 \
+    dnsutils \
+    intel-oneapi-mpi-2021.14 \
     && apt remove -y gnupg2 ca-certificates \
     && apt autoremove -y \
     && rm -rf /var/lib/apt/lists/*