Fix up Ubnuntu server installation

2025-06-23 04:30:05 +00:00
parent 0ac3a94009
commit c3dc444a8f
6 changed files with 82 additions and 66 deletions
--- a/Dockerfile.debian
+++ b/Dockerfile.debian
@@ -307,11 +307,6 @@ RUN \
    && \
    echo

-# TODO: Deprecate neko
-# Install Neko server
-COPY --from=m1k1o/neko:base /usr/bin/neko /usr/bin/neko
-COPY --from=m1k1o/neko:base /var/www /var/www
-
 # Install Web Frontend
 ARG NODE_VERSION="v20"
 ARG FRONTEND_VERSION=a8eb92f
@@ -464,6 +459,28 @@ RUN \
    && \
    echo

+# Install Steam
+RUN \
+    echo "**** Update apt database ****" \
+        && dpkg --add-architecture i386 \
+        && apt-get update \
+    && \
+    echo "**** Install Steam ****" \
+        && apt-get install -y --no-install-recommends \
+            steam-installer \
+            gamescope \
+        && ln -sf /usr/games/steam /usr/bin/steam \
+    && \
+    echo "**** Section cleanup ****" \
+        && apt-get clean autoclean -y \
+        && apt-get autoremove -y \
+        && rm -rf \
+            /var/lib/apt/lists/* \
+            /var/tmp/* \
+            /tmp/* \
+    && \
+    echo
+
 # Install Sunshine
 COPY --from=lizardbyte/sunshine:v2025.122.141614-debian-bookworm /sunshine.deb /usr/src/sunshine.deb
 RUN \
@@ -488,27 +505,10 @@ RUN \
    && \
    echo

-# Install Steam
-RUN \
-    echo "**** Update apt database ****" \
-        && dpkg --add-architecture i386 \
-        && apt-get update \
-    && \
-    echo "**** Install Steam ****" \
-        && apt-get install -y --no-install-recommends \
-            steam-installer \
-            gamescope \
-        && ln -sf /usr/games/steam /usr/bin/steam \
-    && \
-    echo "**** Section cleanup ****" \
-        && apt-get clean autoclean -y \
-        && apt-get autoremove -y \
-        && rm -rf \
-            /var/lib/apt/lists/* \
-            /var/tmp/* \
-            /tmp/* \
-    && \
-    echo
+# TODO: Deprecate neko
+# Install Neko server
+COPY --from=m1k1o/neko:base /usr/bin/neko /usr/bin/neko
+COPY --from=m1k1o/neko:base /var/www /var/www

 # Various other tools
 ARG DUMB_INIT_VERSION=1.2.5
--- a/docs/compose-files/docker-compose.nvidia.privileged.yml
+++ b/docs/compose-files/docker-compose.nvidia.privileged.yml
@@ -15,15 +15,7 @@ services:
        hard: 524288

    # GPU PASSTHROUGH
-    deploy:
-      resources:
-        reservations:
-          # Enable support for NVIDIA GPUs.
-          # 
-          # Ref: https://docs.docker.com/compose/gpu-support/#enabling-gpu-access-to-service-containers
-          devices:
-            - capabilities: [gpu]
-              device_ids: ["${NVIDIA_VISIBLE_DEVICES}"]
+    runtime: nvidia

    # NETWORK:
    ## NOTE:  With this configuration, if we do not use the host network, then physical device input
--- a/docs/compose-files/docker-compose.nvidia.yml
+++ b/docs/compose-files/docker-compose.nvidia.yml
@@ -20,15 +20,7 @@ services:
      - apparmor:unconfined

    # GPU PASSTHROUGH
-    deploy:
-      resources:
-        reservations:
-          # Enable support for NVIDIA GPUs.
-          # 
-          # Ref: https://docs.docker.com/compose/gpu-support/#enabling-gpu-access-to-service-containers
-          devices:
-            - capabilities: [gpu]
-              device_ids: ["${NVIDIA_VISIBLE_DEVICES}"]
+    runtime: nvidia

    # NETWORK:
    ## NOTE:  With this configuration, if we do not use the host network, then physical device input
--- a/docs/ubuntu-server.md
+++ b/docs/ubuntu-server.md
@@ -1,33 +1,57 @@
-# Ubuntu Server
+# Ubuntu Server Setup

-Follow these instructions to install Steam Headless on Ubuntu Server.
+Use these instructions to install **Steam Headless** on an Ubuntu Server system.

-> __Note__
+> ⚠️ **Note**
 >
-> This assumes that your Ubuntu Server has not be configured to run any desktop environment!
-> 
-> This will not work with Ubuntu Desktop.
+> These steps assume you are running a minimal **Ubuntu Server** installation **without any desktop environment**.
+> This setup **will not work** on Ubuntu Desktop.

+---
+
+## INSTALL NVIDIA DRIVER:
+
+Although you're on a server system, using the `-server` variant of the NVIDIA driver can cause compatibility issues.  
+Instead, install the standard driver **without recommended extras**:
+
+```bash
+apt install --no-install-recommends nvidia-driver-570
+```
+
+> 🔍 Feel free to `570` with the latest available version.
+
+To find the latest version of the standard (non-`-server`, non-`-open`) drivers, run:
+
+```bash
+apt-cache search ^nvidia-driver- | awk '{print $1}' | grep -vE '(-server|-open)' | xargs -n1 apt-cache policy | awk '/^nvidia-driver-/{driver=$1} /Candidate:/ {print driver, $2}'
+```
+
+---

 ## INSTALL DOCKER:

-Install docker-ce to your Ubuntu server following the [official instructions](https://docs.docker.com/engine/install/ubuntu/).
+Install `docker-ce` on your Ubuntu server by following the [official Docker instructions](https://docs.docker.com/engine/install/ubuntu/).

-Ensure you install the `docker-compose-plugin` mentioned within these instructions
+Make sure you also install the `docker-compose-plugin` as noted in the Docker documentation.

+---

 ## INSTALL NVIDIA CONTAINER TOOLKIT

-The easiest way to get running with NVIDIA GPUs is to install the [NVIDIA Container Toolkit](https://github.com/NVIDIA/nvidia-container-toolkit?tab=readme-ov-file).
+To enable GPU support inside Docker containers, install the [NVIDIA Container Toolkit](https://github.com/NVIDIA/nvidia-container-toolkit?tab=readme-ov-file).

-Follow the [official instructions](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html#installing-with-apt) for installing the container toolkit to your Ubuntu server with apt.
+Follow the [APT-based installation steps](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html#installing-with-apt) provided in the official documentation.

-If you do this, ensure that when you configure Docker Compose in the next step you choose the `nvidia` runtime.
+Once installed, configure Docker to use the NVIDIA runtime by default:

-Alternately, it is possible to run the container without the NVIDIA runtime by uncommenting the `/dev/nvidia` devices in the Compose file.
+```bash
+sudo nvidia-ctk runtime configure --runtime=docker
+```

+> 💡 You *can* also run the container without the NVIDIA runtime by manually uncommenting the `/dev/nvidia*` device entries in the Compose file — but this approach is **not recommended**.
+
+---

 ## CONFIGURE DOCKER COMPOSE:

-Once you have installed docker, follow the [Compose Files](./docker-compose.md) section and select the right configuration file for your hardware.
-
+After installing Docker, proceed to the [Compose Files](./docker-compose.md) section and select the appropriate configuration for your hardware setup.
--- a/overlay/entrypoint.sh
+++ b/overlay/entrypoint.sh
@@ -42,6 +42,11 @@ function print_error {
    echo -e "\e[31mERROR: ${@}\e[0m"
 }

+function print_note {
+    # Cyan
+    echo -e "\e[36mNOTE: ${@}\e[0m"
+}
+
 # Execute all container init scripts
 for init_script in /etc/cont-init.d/*.sh ; do
    echo
--- a/overlay/etc/cont-init.d/60-configure_gpu_driver.sh
+++ b/overlay/etc/cont-init.d/60-configure_gpu_driver.sh
@@ -41,8 +41,8 @@ function download_driver {
        # Try downloading from a list of NVIDIA driver hosting servers
        stripped_version="${nvidia_host_driver_version#v}" # Strip 'v' if present
        declare -a sources=(
-            "http://download.nvidia.com/XFree86/Linux-x86_64/${nvidia_host_driver_version}/NVIDIA-Linux-x86_64-${nvidia_host_driver_version}.run"
-            "http://us.download.nvidia.com/XFree86/Linux-x86_64/${nvidia_host_driver_version}/NVIDIA-Linux-x86_64-${nvidia_host_driver_version}.run"
+            "https://download.nvidia.com/XFree86/Linux-x86_64/${nvidia_host_driver_version}/NVIDIA-Linux-x86_64-${nvidia_host_driver_version}.run"
+            "https://us.download.nvidia.com/XFree86/Linux-x86_64/${nvidia_host_driver_version}/NVIDIA-Linux-x86_64-${nvidia_host_driver_version}.run"
            "https://github.com/flathub/org.freedesktop.Platform.GL.nvidia/releases/download/cuda/NVIDIA-Linux-x86_64-${stripped_version}.run"
        )

@@ -59,10 +59,11 @@ function download_driver {
            fi
        done

+        print_note "Visit https://download.nvidia.com/XFree86/Linux-x86_64/ in a browser and find the closest match to version '${nvidia_host_driver_version:?}', then set that version in the NVIDIA_DRIVER_VERSION environment variable."
        print_error "Unable to download driver from any source. Exit!"
+        sleep 10
        return 1
    fi
-
 }

 function install_nvidia_driver {
@@ -119,6 +120,8 @@ function install_nvidia_driver {
                --no-install-libglvnd \
                >"${USER_HOME:?}/Downloads/nvidia_gpu_install.log" 2>&1
        fi
+    else
+        print_step_header "NVIDIA driver version ${nvidia_settings_version:-} is already installed"
    fi
 }

@@ -240,15 +243,15 @@ else
    print_header "No AMD device found"
 fi
 # NVIDIA GPU
-if [ "${nvidia_pci_address:-}X" != "X" ]; then
-    print_header "Found NVIDIA device '${nvidia_gpu_name:?}'"
-    install_nvidia_driver
-    patch_nvidia_driver
-elif [ "${NVIDIA_DRIVER_VERSION:-}X" != "X" ]; then
+if [ "${NVIDIA_DRIVER_VERSION:-}X" != "X" ]; then
    export nvidia_host_driver_version="${NVIDIA_DRIVER_VERSION:?}"
    print_header "Forcing install of NVIDIA driver version '${nvidia_host_driver_version:?}' because the 'NVIDIA_DRIVER_VERSION' variable is set."
    install_nvidia_driver
    patch_nvidia_driver
+elif [ "${nvidia_pci_address:-}X" != "X" ]; then
+    print_header "Found NVIDIA device '${nvidia_gpu_name:?}'"
+    install_nvidia_driver
+    patch_nvidia_driver
 else
    print_header "No NVIDIA device found"
 fi