Cloning preparation script for linux systems.

sysprep_linux.sh

#!/bin/bash
# Does the equivalent of sysprep for linux boxes to prepare them for cloning.
# SPDX-License-Identifier: GPL-3.0-or-later

# For issues or updated versions of this script, browse to the following URL:
# https://gist.github.com/AfroThundr3007730/ff5229c5b1f9a018091b14ceac95aa55

set -euo pipefail
shopt -s extdebug nullglob

AUTHOR='AfroThundr'
BASENAME="${0##*/}"
MODIFIED='20241217'
VERSION='1.9.1'

sysprep.parse_args() {
    (($# > 0)) || utils.die 'No arguments specified, use -h for help.'
    while (($# > 0)); do
        case $1 in
        -v) printf '%s: Version %s, updated %s by %s\n' \
            "$BASENAME" "$VERSION" "$MODIFIED" "$AUTHOR" &&
            shift && (($# > 0)) || exit 0 ;;
        -h)
            printf 'Cloning preparation script for linux systems.\n\n'
            printf 'Usage: %s [-v ] (-h | -y [-b] [-l <log_file>] [-s])\n' \
                "$BASENAME"
            printf '\nOptions:\n'
            printf '  -h  Display this help text.\n'
            printf '  -b  Used for firstboot (internal).\n'
            printf '  -l  Specify log file location.\n'
            printf '  -s  Shutdown on completion.\n'
            printf '  -v  Emit version header.\n'
            printf '  -y  Confirm sysprep.\n'
            exit 0
            ;;
        -b) FIRSTBOOT=true && break ;;
        -l) LOGFILE=$2 && shift 2 ;;
        -s) SHUTDOWN=true && shift ;;
        -y) CONFIRM=true && shift ;;
        *) utils.die 'Invalid argument specified, use -h for help.' ;;
        esac
    done
    [[ ${CONFIRM:-} == true ]] || utils.die 'Confirm with -y to start sysprep.'
}

utils.die() { printf '\e[31m%s\n\e[m' "${1:-}" >&2 && exit 1; }

utils.say() {
    LOGFILE=${LOGFILE:=/var/log/sysprep.log}
    [[ ${TZOFFSET:-} ]] || utils.get_tz_offset_secs -s
    [[ -n $LOGFILE && ! $LOGFILE == no ]] && {
        [[ -f $LOGFILE ]] || (umask 027 && : >"$LOGFILE")
        printf '%(%FT%TZ)T: %s\n' $((EPOCHSECONDS - TZOFFSET)) "$1" >>"$LOGFILE"
    }
    printf '%(%FT%TZ)T: %s\n' $((EPOCHSECONDS - TZOFFSET)) "$1"
}

utils.get_tz_offset_secs() {
    local off && printf -v off '\n%(%z)T' -1
    off=${off: -5:1}$((${off: -4:2} * 3600 + ${off: -2:2} * 60))
    [[ $1 == -s ]] || printf '%s\n' "$off" && declare -gi TZOFFSET="$off"
}

sysprep.detect_os() {
    [[ -f /etc/os-release ]] && source /etc/os-release
    if [[ ${ID:-} =~ (fedora|rhel|centos) ||
        ${ID_LIKE:-} =~ (fedora|rhel|centos) ]]; then
        FEDORA_DERIV=true
    elif [[ ${ID:-} =~ (debian|ubuntu|mint) ||
        ${ID_LIKE:-} =~ (debian|ubuntu|mint) ]]; then
        DEBIAN_DERIV=true
    else
        utils.say 'An unknown base linux distribution was detected.'
        utils.say 'This script works with Debian and Fedora based distros.'
        exit 1
    fi
}

sysprep.apt_purge() {
    apt remove -qy --purge "$(
        for i in $(find /boot/vmlinuz-* | sort | head -n -1 |
            grep -v "$(uname -r)" | cut -d- -f2-); do
            echo linux-{image,headers,modules}-"$i"
        done
    )" &>/dev/null && apt autoremove -qy --purge &>/dev/null
}

sysprep.firstboot() {
    utils.say 'Running sysprep first-boot setup script.'
    [[ ${DEBIAN_DERIV:-} == true ]] && {
        find /etc/ssh/*key &>/dev/null || {
            utils.say 'Regenerating SSH host keys...'
            dpkg-reconfigure openssh-server
        }
    }
    [[ ${HOSTNAME:-} == CHANGEME ]] && {
        utils.say 'Regenerating hostname and rebooting...'
        hostnamectl set-hostname \
            "LINUX-$(tr -cd '[:upper:][:digit:]' </dev/urandom | head -c 9)"
        systemctl reboot
    }
    [[ -f /var/lib/aide/aide.db.gz ]] && {
        utils.say 'Regenerating AIDE database...'
        aide --update && mv -f /var/lib/aide/aide.db{.new,}.gz
    }
    utils.say 'Sysprep firtst-boot setup complete, disabling service.'
    systemctl -q disable sysprep-firstboot
    exit 0
}

sysprep.clean_packages() {
    utils.say 'Removing old kernels.'
    if [[ ${FEDORA_DERIV:-} == true ]]; then
        if command -v dnf &>/dev/null; then
            dnf remove -qy "$(dnf repoquery --installonly --latest-limit=-1)"
        else
            command -v package-cleanup &>/dev/null ||
                yum install -qy yum-utils &>/dev/null
            package-cleanup -qy --oldkernels --count=1 &>/dev/null
        fi
    elif [[ ${DEBIAN_DERIV:-} == true ]]; then
        command -v purge-old-kernels &>/dev/null ||
            apt install -qy byobu &>/dev/null
        purge-old-kernels -qy --keep 1 &>/dev/null || sysprep.apt_purge
    fi
    utils.say 'Clearing package cache.'
    if [[ ${FEDORA_DERIV:-} == true ]]; then
        yum clean all -q &>/dev/null
        rm -rf /var/cache/yum/*
    elif [[ ${DEBIAN_DERIV:-} == true ]]; then
        apt clean &>/dev/null
        rm -rf /var/cache/apt/archives/*
    fi
}

sysprep.clean_logs() {
    utils.say 'Clearing old logs.'
    logrotate -f /etc/logrotate.conf &>/dev/null || :
    find /var/log -type f -regextype posix-extended -regex \
        ".*/*(-[0-9]{8}|.[0-9]|.gz)$" -delete
    command -V setenforce &>/dev/null && setenforce 0
    journalctl -q --sync --rotate --vacuum-files 1
    rm -f /var/log/dmesg.old /var/log/anaconda/*
    utils.say 'Clearing audit logs.'
    : >/var/log/audit/audit.log
    : >/var/log/wtmp
    : >/var/log/lastlog
    : >/var/log/grubby
}

sysprep.clean_network() {
    utils.say 'Clearing udev persistent rules.'
    rm -f /etc/udev/rules.d/70*
    utils.say 'Removing MACs/UUIDs from network configs.'
    [[ ${FEDORA_DERIV:-} == true ]] && {
        for file in /etc/sysconfig/network-scripts/ifcfg-*; do
            sed -ri '/^(HWADDR|UUID)=/d' "$file"
        done
    }
    for file in /etc/NetworkManager/system-connections/*; do
        sed -ri '/^(mac-address|uuid)=/d' "$file"
    done
}

sysprep.clean_files() {
    utils.say 'Cleaning out temp directories.'
    rm -rf /tmp/* /var/tmp/* /var/cache/*
    utils.say 'Cleaning up root home directory.'
    : >"${HISTFILE:-/dev/null}" && unset HISTFILE
    rm -f /root/anaconda-ks.cfg /root/.ssh/* /root/.gnupg/*
}

sysprep.generalize() {
    utils.say 'Removing SSH host keys.'
    rm -f /etc/ssh/*key*
    utils.say 'Clearing machine-id'
    : >/etc/machine-id
    utils.say 'Removing random-seed'
    rm -f /var/lib/systemd/random-seed
    [[ -f /opt/McAfee/agent/bin/maconfig ]] && {
        utils.say 'Resetting McAfee Agent'
        systemctl stop mcafee.ma
        command -V setenforce &>/dev/null && setenforce 0
        /opt/McAfee/agent/bin/maconfig -enforce -noguid
    }
    utils.say 'Resetting hostname.'
    hostnamectl set-hostname 'CHANGEME'
}

sysprep.setup_firstboot() {
    utils.say 'Enabling sysprep firstboot service.'
    local _unit=/etc/systemd/system/sysprep-firstboot.service
    [[ -f $_unit ]] || cat <<'EOF' >$_unit
[Unit]
Description=Sysprep first-boot setup tasks
[Service]
Type=simple
ExecStart=/usr/local/sbin/sysprep -y -b
[Install]
WantedBy=multi-user.target
EOF
    systemctl -q enable sysprep-firstboot
}

sysprep.run() {
    sysprep.parse_args "$@"
    sysprep.detect_os
    utils.say 'Beginning sysprep.'
    [[ ${FIRSTBOOT:-} == true ]] && sysprep.firstboot
    utils.say 'Stopping logging and auditing daemons.'
    systemctl -q stop rsyslog.service
    service auditd stop &>/dev/null
    sysprep.clean_packages
    sysprep.clean_logs
    sysprep.clean_network
    sysprep.clean_files
    sysprep.generalize
    sysprep.setup_firstboot
    utils.say 'End of sysprep.'
    [[ ${SHUTDOWN:-} == true ]] &&
        utils.say 'Shutting down the system.' && systemctl poweroff
}

# Only execute if not being sourced
[[ ${BASH_SOURCE[0]} == "$0" ]] || return 0 && sysprep.run "$@"

@boktai1000 I will admit that my choice of uppercasing the hostname was mostly for cosmetic reasons. The randomly generated hostname is similar to what a new Windows box will end up with after a sysprep, so I just copied that behavior.

As for software breaking on an uppercase name, the RFC doesn't actually prohibit an uppercase name, so they are still technically valid. It would seem that that particular program is making an overly strict assumption on what constitutes a valid hostname. Most implementations simply treat hostnames case-insensitively to avoid the issue, so 'LINUX' and 'linux' would be the same name.

I think I'll keep this particular quirk in for now (purely for cosmetic reasons at the moment), but if you find more info on why hostnames really should be lowercase at all times, or find more packages that break because of this, I'd be interested to know.

@AfroThundr3007730 note to self: check out https://github.com/ferricoxide/AMIgen7/blob/master/CleanChroot.sh