fix: skip fabric manager on arm64 (Grace-Blackwell GB200/GB300)

arm64 cuda nodes are the Grace-Blackwell SKUs (GB200/GB300), which have no
node-local fabric manager -- the NVLink fabric is managed out-of-band via IMEX.
The driver container should neither install nor bundle fabricmanager there.

Today device_init() runs the fabricmanager package installer for every cuda
node. On arm64 it computes NVIDIA_FM_ARCH from $ARCH (uname -m = "aarch64")
and then only remaps to "sbsa" when the value equals "arm64" -- which never
matches -- so it looks for /opt/gpu/fabricmanager-linux-aarch64-<ver>/... which
does not exist (the tarball download.sh bundles is named "sbsa"). The installer
exits 127; installGPUDriverImage retries the whole container until CSE hits its
~780s timeout and exits 84, so ensureKubelet never runs and the node never joins
the cluster. Surfaced by the GB300 managed-driver (--gpu-driver Install) e2e.

Fix: gate fabric manager to x86_64 only.
- install.sh: run fm_run_package_installer.sh only when ARCH == x86_64; drop the
  dead arm64->sbsa remap.
- download.sh: download/bundle the fabricmanager tarball only when
  TARGETARCH == amd64, so the arm64 image no longer ships an unused tarball.

Author: xuexu6666

download.sh

@@ -48,8 +48,11 @@ install_fabric_manager () {
     mv /opt/gpu/fm_run_package_installer.sh /opt/gpu/fabricmanager-linux-${NVIDIA_FM_ARCH}-${DRIVER_VERSION}/sbin/fm_run_package_installer.sh
 }
 
-if [[ "${DRIVER_KIND}" == "cuda" ]]; then
-    # download fabricmanager for nvlink based systems, e.g. multi instance gpu vms.
+if [[ "${DRIVER_KIND}" == "cuda" && "${TARGETARCH}" == "amd64" ]]; then
+    # download fabricmanager for nvlink based systems, e.g. multi instance gpu vms (x86_64 only).
+    # arm64 cuda nodes are the Grace-Blackwell SKUs (GB200/GB300), which have no node-local
+    # fabric manager -- their NVLink fabric is managed out-of-band via IMEX -- so we neither
+    # install (see install.sh device_init) nor bundle it here.
    install_fabric_manager
 fi
 

install.sh

@@ -159,14 +159,14 @@ device_init() {
     cp -r /usr/bin/lib64/lib64/* "/usr/lib/${ARCH}-linux-gnu/"
     nvidia-smi
 
-    # install fabricmanager for nvlink based systems
-    if [[ "${DRIVER_KIND}" == "cuda" ]]; then
-        NVIDIA_FM_ARCH=$ARCH
-        if [ "$NVIDIA_FM_ARCH" = "arm64" ]; then
-            # NVIDIA uses the name "SBSA" for ARM64 platforms for the fabric manager. See https://en.wikipedia.org/wiki/Server_Base_System_Architecture
-            NVIDIA_FM_ARCH="sbsa"
-        fi
-        bash /opt/gpu/fabricmanager-linux-${NVIDIA_FM_ARCH}-${DRIVER_VERSION}/sbin/fm_run_package_installer.sh
+    # install fabricmanager for nvlink based systems (x86_64 only).
+    # arm64 cuda nodes are the Grace-Blackwell SKUs (GB200/GB300), which have NO
+    # node-local fabric manager -- their NVLink fabric is managed out-of-band via
+    # IMEX. Running fm_run_package_installer.sh there is both wrong and fails (the
+    # bundled tarball is named "sbsa", not uname -m's "aarch64"), stalling CSE past
+    # its timeout so the node never starts kubelet. So gate FM to x86_64.
+    if [[ "${DRIVER_KIND}" == "cuda" && "${ARCH}" == "x86_64" ]]; then
+        bash /opt/gpu/fabricmanager-linux-${ARCH}-${DRIVER_VERSION}/sbin/fm_run_package_installer.sh
     fi
 
     mkdir -p /etc/containerd/config.d