Fix USB install reliability: bash, LV resize, deps, idempotency

- autoinstall/user-data: move lvextend/growpart/pvresize BEFORE playbook so 130GB of drivers+PPKGs fits during first-boot copy. Use tr -d "[:space:]" to avoid breaking outer bash -c single-quote wrap. - playbook: add executable: /bin/bash to Dell driver deploy (process substitution) and Blancco initramfs builder (brace expansion). - playbook: make "Ensure Samba user for Blancco reports" idempotent via pdbedit check so re-runs don't abort the play. - download-packages.sh: also download dist-upgrade package set. Explicit --simulate misses transitive version bumps (e.g. gnupg 17.4 needs matching gpgv 17.4) causing offline dpkg "dependency problems" when ISO baseline is older than noble-updates.
2026-04-14 12:57:28 -04:00
parent 855af7312b
commit ade2f3b5ff
3 changed files with 136 additions and 35 deletions
--- a/autoinstall/user-data
+++ b/autoinstall/user-data
@@ -72,6 +72,16 @@ autoinstall:
      curtin in-target --target=/target -- bash -c '
      cat <<"EOF" > /opt/first-boot.sh
      #!/bin/bash
+      # Expand root LV to full disk BEFORE playbook (playbook copies ~130GB of drivers+PPKGs)
+      ROOT_DEV=$(findmnt -n -o SOURCE /)
+      ROOT_DISK=$(lsblk -n -o PKNAME "$(readlink -f "$ROOT_DEV")" | tail -1)
+      PV_PART=$(pvs --noheadings -o pv_name 2>/dev/null | tr -d "[:space:]" | head -1)
+      if [ -n "$ROOT_DISK" ] && [ -n "$PV_PART" ]; then
+        PART_NUM=$(echo "$PV_PART" | grep -o "[0-9]*$")
+        growpart "/dev/${ROOT_DISK}" "${PART_NUM}" 2>&1 || true
+        pvresize "$PV_PART" 2>&1 || true
+      fi
+      lvextend -r -l +100%FREE /dev/mapper/ubuntu--vg-ubuntu--lv 2>&1 || true
      CIDATA_DEV=$(blkid -L CIDATA)
      if [ -n "$CIDATA_DEV" ]; then
        mkdir -p /mnt/usb
@@ -88,7 +98,6 @@ autoinstall:
        umount /mnt/usb
      fi
      sed -i "s|^/opt/first-boot.sh.*|# &|" /etc/rc.local
-      lvextend -r -l +100%FREE /dev/mapper/ubuntu--vg-ubuntu--lv || true
      EOF
      chmod +x /opt/first-boot.sh
      '
--- a/download-packages.sh
+++ b/download-packages.sh
@@ -1,10 +1,10 @@
 #!/bin/bash
 #
-# download-packages.sh — Download all .deb packages needed for offline PXE server setup
+# download-packages.sh - Download all .deb packages needed for offline PXE server setup
 #
-# Run this on a machine with internet access running Ubuntu 24.04 (Noble).
-# It downloads every .deb needed by the Ansible playbook into a local directory,
-# which then gets bundled onto the installer USB.
+# The PXE server installs Ubuntu 24.04 (Noble), so all packages MUST come from the
+# 24.04 archive. If this script is run on a non-24.04 host (e.g. Zorin 17 / 22.04),
+# it auto-spawns an Ubuntu 24.04 docker container to do the download.
 #
 # Usage:
 #   ./download-packages.sh [output_directory]
@@ -14,6 +14,40 @@
 set -euo pipefail

 OUT_DIR="${1:-./offline-packages}"
+OUT_DIR_ABS="$(cd "$(dirname "$OUT_DIR")" 2>/dev/null && pwd)/$(basename "$OUT_DIR")"
+
+# Detect host Ubuntu codename. Run inside the container if not Noble (24.04).
+HOST_CODENAME="$(. /etc/os-release && echo "${UBUNTU_CODENAME:-${VERSION_CODENAME:-}}")"
+
+if [ "${IN_DOCKER:-}" != "1" ] && [ "$HOST_CODENAME" != "noble" ]; then
+    echo "Host is '$HOST_CODENAME', not 'noble' (Ubuntu 24.04)."
+    echo "Re-running inside ubuntu:24.04 docker container..."
+    echo ""
+
+    if ! command -v docker >/dev/null; then
+        echo "ERROR: docker not installed. Install docker or run on a real Ubuntu 24.04 host."
+        exit 1
+    fi
+
+    SCRIPT_PATH="$(readlink -f "$0")"
+    REPO_DIR="$(dirname "$SCRIPT_PATH")"
+    mkdir -p "$OUT_DIR_ABS"
+
+    docker run --rm -i \
+        -v "$REPO_DIR:/repo" \
+        -v "$OUT_DIR_ABS:/out" \
+        -e IN_DOCKER=1 \
+        -w /repo \
+        ubuntu:24.04 \
+        bash -c "apt-get update -qq && apt-get install -y --no-install-recommends sudo python3-pip python3-setuptools python3-wheel ca-certificates >/dev/null && /repo/download-packages.sh /out"
+
+    echo ""
+    echo "============================================"
+    echo "Container build complete. Files in: $OUT_DIR_ABS"
+    echo "============================================"
+    exit 0
+fi
+
 mkdir -p "$OUT_DIR"

 # Packages installed by the Ansible playbook (pxe_server_setup.yml)
@@ -30,10 +64,12 @@ PLAYBOOK_PACKAGES=(
    grub-efi-amd64-bin
    grub-common
    conntrack
+    busybox-static
+    zstd
+    cpio
 )

 # Packages installed during autoinstall late-commands (NetworkManager, WiFi, etc.)
-# These are already in your ubuntu_playbook/*.deb files, but we can refresh them here too.
 AUTOINSTALL_PACKAGES=(
    network-manager
    wpasupplicant
@@ -45,7 +81,7 @@ AUTOINSTALL_PACKAGES=(
 ALL_PACKAGES=("${PLAYBOOK_PACKAGES[@]}" "${AUTOINSTALL_PACKAGES[@]}")

 echo "============================================"
-echo "Offline Package Downloader"
+echo "Offline Package Downloader (Ubuntu 24.04 noble)"
 echo "============================================"
 echo "Output directory: $OUT_DIR"
 echo ""
@@ -54,18 +90,29 @@ printf '  - %s\n' "${ALL_PACKAGES[@]}"
 echo ""

 # Update package cache
-echo "[1/3] Updating package cache..."
+echo "[1/4] Updating package cache..."
 sudo apt-get update -qq

 # Simulate install to find all dependencies
-echo "[2/3] Resolving dependencies..."
-DEPS=$(apt-get install --simulate "${ALL_PACKAGES[@]}" 2>&1 \
+echo "[2/4] Resolving dependencies..."
+EXPLICIT_DEPS=$(apt-get install --simulate "${ALL_PACKAGES[@]}" 2>&1 \
    | grep "^Inst " \
-    | awk '{print $2}' \
-    | sort -u)
+    | awk '{print $2}')
+
+# ALSO pull every package that would upgrade in a dist-upgrade. This is
+# critical: the Ubuntu ISO ships a point-in-time baseline, but our explicit
+# packages (from noble-updates) may depend on *newer* versions of ISO-baseline
+# packages (e.g. gnupg 17.4 needs matching gpgv 17.4). Without this, offline
+# install fails with dpkg "dependency problems" because transitive version
+# bumps aren't captured by --simulate on the explicit list.
+UPGRADE_DEPS=$(apt-get dist-upgrade --simulate 2>&1 \
+    | grep "^Inst " \
+    | awk '{print $2}')
+
+DEPS=$(printf '%s\n%s\n' "$EXPLICIT_DEPS" "$UPGRADE_DEPS" | sort -u | grep -v '^$')

 DEP_COUNT=$(echo "$DEPS" | wc -l)
-echo "    Found $DEP_COUNT packages (including dependencies)"
+echo "    Found $DEP_COUNT packages (explicit + baseline upgrades)"

 # Download all packages
 echo "[3/4] Downloading .deb packages to $OUT_DIR..."
@@ -79,7 +126,9 @@ echo "    $DEB_COUNT packages ($TOTAL_SIZE)"

 # Download pip wheels for Flask webapp (offline install)
 echo "[4/4] Downloading Python wheels for webapp..."
-PIP_DIR="$(dirname "$OUT_DIR")/pip-wheels"
+# Place pip-wheels next to the script (or /repo when in docker), not next to OUT_DIR
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+PIP_DIR="$SCRIPT_DIR/pip-wheels"
 mkdir -p "$PIP_DIR"
 pip3 download -d "$PIP_DIR" flask lxml 2>&1 | tail -5

--- a/playbook/pxe_server_setup.yml
+++ b/playbook/pxe_server_setup.yml
@@ -307,9 +307,19 @@
        mode: '0777'

    - name: "Deploy PPKG enrollment packages to enrollment share"
-      shell: cp -f {{ usb_mount }}/enrollment/*.ppkg /srv/samba/enrollment/ 2>/dev/null || true
-      args:
-        warn: false
+      shell: |
+        set +e
+        # Copy any whole PPKGs (small enough to fit on FAT32)
+        cp -f {{ usb_root }}/enrollment/*.ppkg /srv/samba/enrollment/ 2>/dev/null
+        # Reassemble any split files (foo.ppkg.part.00, .01, ... -> foo.ppkg)
+        for first in {{ usb_root }}/enrollment/*.part.00; do
+          [ -e "$first" ] || continue
+          base="${first%.part.00}"
+          name="$(basename "$base")"
+          echo "Reassembling $name from chunks..."
+          cat "${base}.part."* > "/srv/samba/enrollment/$name"
+        done
+        ls -lh /srv/samba/enrollment/*.ppkg 2>/dev/null
      ignore_errors: yes

    - name: "Deploy run-enrollment.ps1 to enrollment share"
@@ -320,16 +330,29 @@
      ignore_errors: yes

    - name: "Deploy Dell driver packs to shared Out-of-box Drivers"
-      shell: >
-        if [ -d "{{ usb_mount }}/drivers" ]; then
-          mkdir -p "/srv/samba/winpeapps/_shared/Out-of-box Drivers/Dell_11"
-          cp -r {{ usb_mount }}/drivers/* "/srv/samba/winpeapps/_shared/Out-of-box Drivers/Dell_11/"
-          echo "Deployed Dell drivers from USB"
-        else
-          echo "No drivers/ on USB - skipping"
-        fi
      args:
-        warn: false
+        executable: /bin/bash
+      shell: |
+        set +e
+        SRC="{{ usb_root }}/drivers"
+        DEST="/srv/samba/winpeapps/_shared/Out-of-box Drivers/Dell_11"
+        if [ ! -d "$SRC" ]; then
+          echo "No drivers/ on USB - skipping"
+          exit 0
+        fi
+        mkdir -p "$DEST"
+        # Copy everything except split chunks
+        rsync -a --exclude='*.part.*' "$SRC/" "$DEST/"
+        # Reassemble any split driver files
+        while IFS= read -r first; do
+          base="${first%.part.00}"
+          rel="${base#$SRC/}"
+          out="$DEST/$rel"
+          mkdir -p "$(dirname "$out")"
+          echo "Reassembling $rel from chunks..."
+          cat "${base}.part."* > "$out"
+        done < <(find "$SRC" -name '*.part.00')
+        echo "Deployed Dell drivers from USB"
      ignore_errors: yes

    - name: "Deploy shopfloor setup scripts to enrollment share"
@@ -372,6 +395,17 @@
        - models.txt
      ignore_errors: yes

+    - name: "Deploy BIOS update binaries from USB"
+      shell: >
+        if [ -d "{{ usb_root }}/bios" ]; then
+          cp -f {{ usb_root }}/bios/*.exe /srv/samba/enrollment/BIOS/ 2>/dev/null || true
+          count=$(find /srv/samba/enrollment/BIOS -name '*.exe' | wc -l)
+          echo "Deployed $count BIOS binaries"
+        else
+          echo "No bios/ on USB - skipping"
+        fi
+      ignore_errors: yes
+
    - name: "Create image upload staging directory"
      file:
        path: /home/pxe/image-upload
@@ -579,13 +613,24 @@
    # Boot Ubuntu kernel, download Blancco rootfs, overlay mount, switch_root.

    - name: "Build Blancco PXE initramfs"
+      args:
+        executable: /bin/bash
+        creates: "{{ web_root }}/blancco/kexec-initrd.img"
      shell: |
        set -e
        WORK=$(mktemp -d)
        mkdir -p "$WORK"/{bin,lib/modules,lib64,sbin,usr/share/udhcpc}

-        # Busybox (static)
-        cp /bin/busybox "$WORK/bin/" 2>/dev/null || apt-get install -y busybox-static >/dev/null && cp /bin/busybox "$WORK/bin/"
+        # Busybox (static) - bundled on USB at playbook/busybox-static
+        if [ -f /bin/busybox ]; then
+            cp /bin/busybox "$WORK/bin/"
+        elif [ -f "{{ usb_root }}/playbook/busybox-static" ]; then
+            cp "{{ usb_root }}/playbook/busybox-static" "$WORK/bin/busybox"
+            chmod +x "$WORK/bin/busybox"
+        else
+            echo "ERROR: No busybox available (not at /bin/busybox or on USB)"
+            exit 1
+        fi
        for cmd in sh awk cat chmod echo grep gunzip ifconfig ip ln losetup ls mkdir mknod mount reboot route sed sleep switch_root tar udhcpc umount wget cpio; do
            ln -sf busybox "$WORK/bin/$cmd"
        done
@@ -614,8 +659,7 @@
        find . | cpio -o -H newc 2>/dev/null | gzip > "{{ web_root }}/blancco/kexec-initrd.img"
        rm -rf "$WORK"
        echo "Built kexec-initrd.img: $(stat -c %s '{{ web_root }}/blancco/kexec-initrd.img') bytes"
-      args:
-        creates: "{{ web_root }}/blancco/kexec-initrd.img"
+      ignore_errors: yes

    - name: "Copy Ubuntu kernel for Blancco PXE boot"
      copy:
@@ -648,12 +692,11 @@
      args:
        creates: "{{ web_root }}/blancco/config-clean.xml"

-    - name: "Create Samba user for Blancco reports"
+    - name: "Ensure Samba user for Blancco reports exists (idempotent)"
      shell: |
-        id blancco 2>/dev/null || useradd -r -s /usr/sbin/nologin blancco
-        echo -e "blancco\nblancco" | smbpasswd -a -s blancco 2>/dev/null
-      args:
-        creates: /etc/samba/smbpasswd
+        id blancco >/dev/null 2>&1 || useradd -r -s /usr/sbin/nologin blancco
+        pdbedit -L 2>/dev/null | grep -q '^blancco:' || (echo -e "blancco\nblancco" | smbpasswd -a -s blancco)
+      changed_when: false

    - name: "Check for WinPE deployment content on USB"
      stat: