Merge pull request #284 from outis1one/claude/ionos-script-integration-x32ofw

Claude/ionos script integration x32ofw
This commit is contained in:
Outis
2026-08-10 14:57:06 -04:00
committed by GitHub
6 changed files with 206 additions and 23 deletions
+28
View File
@@ -299,6 +299,34 @@ docker compose pull && docker compose up -d # update
docker compose down # stop
```
## SSH key import (GitHub/Launchpad) and disabling password login
Imports your public keys from GitHub and/or Launchpad (Canonical/Ubuntu's
own code-hosting platform) into `~/.ssh/authorized_keys` via
[`ssh-import-id`](https://manpages.ubuntu.com/manpages/noble/man1/ssh-import-id.1.html),
so you can log in with a key instead of a password — then optionally locks
password auth off entirely once at least one key is confirmed imported.
Only your **public** key is ever involved — the same information already
visible on `github.com/<user>.keys`, fetched over HTTPS. No private key
material leaves wherever it was generated, and importing a key does **not**
give this box any ability to authenticate *outward* as you (e.g. it still
can't clone your private repos) — it only grants *inbound* login to
whoever holds the matching private key.
Two ways to run it:
- **During `base` install** — runs automatically as part of the required
setup on a fresh box, right after the SSH server itself is configured
- **Any time** — `sudo ./setup.sh ssh-key-import` runs just this step on
its own: import more keys later (a new admin, a different box), or set
it up on a box that only needs this and nothing else `base` does — e.g.
the home box side of [`vpn-data-mount`](#services)
Password authentication is only offered to be disabled if at least one key
import actually succeeded in that run — never blindly, so you can't get
locked out by declining every import prompt.
## SSH Host aliases
`~/.ssh/config` lets you `ssh <alias>` instead of typing `ssh user@1.2.3.4`
+13
View File
@@ -279,6 +279,19 @@ install_audiobookshelf() {
# CLAUDE.md's "Port collision avoidance" section.
find_free_port WEB_PORT "$WEB_PORT"
# Reset before the chain call, not after — VDM_LAST_MOUNT_POINT is a
# plain global set by services/vpn-data-mount.sh, so without this an
# unrelated earlier vpn-data-mount run in the same setup.sh session
# would silently leak its mount point in here as the default even if
# the user declines below or this chain never runs at all.
unset VDM_LAST_MOUNT_POINT
if declare -F install_vpn-data-mount >/dev/null 2>&1; then
local USE_VPN_DATA=""
prompt_yn "Are your audiobooks on a VPN-connected home box that isn't mounted yet? (y/n):" "n" USE_VPN_DATA
[[ "$USE_VPN_DATA" =~ ^[Yy]$ ]] && install_vpn-data-mount
fi
[ -n "${VDM_LAST_MOUNT_POINT:-}" ] && DEFAULT_AUDIOBOOKS="$VDM_LAST_MOUNT_POINT"
local AUDIOBOOKS_PATH=""
prompt_text "Path to audiobooks folder [$DEFAULT_AUDIOBOOKS]:" "$DEFAULT_AUDIOBOOKS" AUDIOBOOKS_PATH
AUDIOBOOKS_PATH="${AUDIOBOOKS_PATH/#\~/$ACTUAL_HOME}"; AUDIOBOOKS_PATH="${AUDIOBOOKS_PATH%/}"
+13
View File
@@ -304,6 +304,19 @@ install_emby() {
FOLDER_PROMPT="Path to music folder"
fi
# Reset before the chain call, not after — VDM_LAST_MOUNT_POINT is a
# plain global set by services/vpn-data-mount.sh, so without this an
# unrelated earlier vpn-data-mount run in the same setup.sh session
# would silently leak its mount point in here as the default even if
# the user declines below or this chain never runs at all.
unset VDM_LAST_MOUNT_POINT
if declare -F install_vpn-data-mount >/dev/null 2>&1; then
local USE_VPN_DATA=""
prompt_yn "Is your media on a VPN-connected home box that isn't mounted yet? (y/n):" "n" USE_VPN_DATA
[[ "$USE_VPN_DATA" =~ ^[Yy]$ ]] && install_vpn-data-mount
fi
[ -n "${VDM_LAST_MOUNT_POINT:-}" ] && DEFAULT_FOLDER="$VDM_LAST_MOUNT_POINT"
local MEDIA_PATH=""
prompt_text "$FOLDER_PROMPT [$DEFAULT_FOLDER]:" "$DEFAULT_FOLDER" MEDIA_PATH
MEDIA_PATH="${MEDIA_PATH/#\~/$ACTUAL_HOME}"; MEDIA_PATH="${MEDIA_PATH%/}"
+14 -2
View File
@@ -277,8 +277,20 @@ install_filebrowser() {
ensure_docker_dir_ownership "$FB_DIR"
cd "$FB_DIR" || return 1
local FB_PATH=""
prompt_text "Primary files directory to browse [default: $ACTUAL_HOME]:" "$ACTUAL_HOME" FB_PATH
# Reset before the chain call, not after — VDM_LAST_MOUNT_POINT is a
# plain global set by services/vpn-data-mount.sh, so without this an
# unrelated earlier vpn-data-mount run in the same setup.sh session
# would silently leak its mount point in here as the default even if
# the user declines below or this chain never runs at all.
unset VDM_LAST_MOUNT_POINT
if declare -F install_vpn-data-mount >/dev/null 2>&1; then
local USE_VPN_DATA=""
prompt_yn "Is the data you want to browse on a VPN-connected home box that isn't mounted yet? (y/n):" "n" USE_VPN_DATA
[[ "$USE_VPN_DATA" =~ ^[Yy]$ ]] && install_vpn-data-mount
fi
local FB_PATH="" _default_fb_path="${VDM_LAST_MOUNT_POINT:-$ACTUAL_HOME}"
prompt_text "Primary files directory to browse [default: $_default_fb_path]:" "$_default_fb_path" FB_PATH
# Mirrors configure_caddy_for_service's own mode resolution (lib/common.sh):
# explicit CADDY_MODE from the site config wins, then a local ~/docker/caddy,
+130 -18
View File
@@ -29,11 +29,20 @@
# redundant overhead for no added security, and it's the slowest and least
# robust (FUSE reconnect quirks) of the three for an always-on mount.
#
# The share is guest-accessible (no separate Samba username/password to
# manage) because the VPN is the actual access control here — only
# NetBird-connected peers can reach the home box's NetBird IP at all, so a
# second credential layer on top of that doesn't add real security, just
# more secrets to lose track of.
# Each share gets a dedicated Samba account (same username as the SSH user,
# since that Unix account already exists on the home box — no extra remote
# user to create) with its own generated password, stored locally in a
# root-only credentials file, same convention tools/mount-network-drive.sh
# already uses. An earlier version of this made the share guest-accessible
# instead, reasoning the VPN itself was enough access control — reversed
# per direct request (real per-share accounts, not root/guest) and because
# it incidentally fixes a real bug: plain `guest` CIFS mounts with no
# explicit `sec=` can hit "mount error(79): Can not access a needed shared
# library" — a misleadingly-worded cifs-utils message for errno 79
# (ENOKEY), a known rough edge in the kernel cifs.ko keyring/upcall path
# for anonymous sessions specifically, not an actual missing library.
# Credentialed mounts with an explicit sec= take the normal NTLMSSP auth
# path instead and don't hit it.
# ── Standalone bootstrap ──────────────────────────────────────────────────────
if [[ "${BASH_SOURCE[0]}" == "${0}" ]]; then
@@ -95,6 +104,52 @@ _vdm_list_existing() {
fi
}
# ── Name a raw IP so it can be used everywhere instead of typing it again ──
# Deliberately /etc/hosts, not ~/.ssh/config: an SSH Host alias only helps
# the `ssh` command itself resolve a name — mount.cifs (and everything
# else) never consults ~/.ssh/config at all, so an alias alone wouldn't let
# the actual CIFS mount address use a name. /etc/hosts is the one mechanism
# that makes a name resolve for both, which is what "use a name instead of
# the IP" actually needs end to end.
# Sets RESOLVED_HOST (not local — same out-param convention as elsewhere).
_vdm_resolve_host() {
local input="$1" default_name="$2"
RESOLVED_HOST="$input"
# Not a raw IP (already a name, whether from /etc/hosts, real DNS, or
# just typed that way) — nothing to do.
[[ "$input" =~ ^[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+$ ]] || return 0
# Already named from an earlier vpn-data-mount run against this same IP?
local existing
existing="$(awk -v ip="$input" '$1==ip && /# vpn-data-mount/ {print $2; exit}' /etc/hosts 2>/dev/null)"
if [ -n "$existing" ]; then
log_info "Already named '$existing' in /etc/hosts from an earlier mount — using that."
RESOLVED_HOST="$existing"
return 0
fi
# prompt_text always falls back to the given default on blank input
# (same convention every other prompt in this repo relies on) — so this
# always ends up naming the host to something, never truly "blank" for
# "keep using the IP". That's fine: $default_name is already a sensible
# suggestion (the mount's own label), so Enter alone gives a reasonable
# name rather than requiring the extra step of confirming one.
local NAME=""
prompt_text " Name for this home box (used instead of the IP from now on):" "$default_name" NAME
NAME="$(echo "$NAME" | tr -cs 'a-zA-Z0-9-' '-' | sed 's/^-*//;s/-*$//')"
[ -z "$NAME" ] && return 0
if grep -qE "^\S+[[:space:]]+${NAME}([[:space:]]|\$)" /etc/hosts 2>/dev/null; then
log_warning "'$NAME' is already used for a different address in /etc/hosts — keeping the IP instead."
return 0
fi
echo "${input} ${NAME} # vpn-data-mount" >> /etc/hosts
log_success "Added to /etc/hosts: $NAME -> $input (works for SSH, this mount, and anything else on this box)"
RESOLVED_HOST="$NAME"
}
# ── SSH trust: test first, only bootstrap if actually needed ──────────────
# Covers "the home box and VPS already share a key via GitHub import (or any
# other means)" for free — if it already works, nothing below runs at all.
@@ -173,8 +228,12 @@ _vdm_ensure_ssh_trust() {
}
# ── Remote Samba setup, driven entirely over the SSH trust above ──────────
# Sets SMB_PASS (not local — read by the caller to write the local
# credentials file), same out-param convention as lib/common.sh's
# ensure_coturn_user/configure_caddy_for_service.
_vdm_setup_remote_samba() {
local user="$1" host="$2" remote_path="$3" share_name="$4"
SMB_PASS=""
log_info "Checking Samba on the home box..."
# Every ssh call below runs as $ACTUAL_USER, same reason as _vdm_ssh_works.
@@ -186,9 +245,22 @@ _vdm_setup_remote_samba() {
log_success "Samba already installed on the home box."
fi
log_info "Configuring the share on the home box..."
SMB_PASS="$(openssl rand -base64 16 | tr -dc 'a-zA-Z0-9' | head -c 16)"
log_info "Configuring the share and Samba account on the home box..."
# Idempotent: drop any prior block for this exact share name, then
# append a fresh one. Guest-accessible — see the file header for why.
# append a fresh one. Dedicated account (not guest) — see the file
# header for why, and for the errno-79 connection.
#
# The Samba account reuses $user — that Unix account already exists on
# the home box (it's who we're SSH'd in as), so no extra remote user
# needs creating. smbpasswd's password ends up briefly visible in the
# home box's own `ps` output for this one remote command's duration
# (embedded in the command string, not piped — piping through a `-t`
# pty session for a command that may also need an interactive sudo
# password gets stdin-ordering-fragile fast). Accepted tradeoff, same
# spirit as this whole feature already trusting the VPN's reachability
# boundary — see file header.
local remote_cmd
remote_cmd=$(cat << REMOTECMD
set -e
@@ -201,15 +273,18 @@ sudo sed -i "/^\\[${share_name}\\]\$/,/^\$/d" /etc/samba/smb.conf
echo " path = ${remote_path}"
echo " browseable = yes"
echo " read only = no"
echo " guest ok = yes"
echo " force user = \$(whoami)"
echo " guest ok = no"
echo " valid users = ${user}"
echo " force user = ${user}"
} | sudo tee -a /etc/samba/smb.conf >/dev/null
printf '%s\n%s\n' '${SMB_PASS}' '${SMB_PASS}' | sudo smbpasswd -a -s '${user}'
sudo smbpasswd -e '${user}'
sudo systemctl restart smbd
command -v ufw >/dev/null 2>&1 && sudo ufw allow samba >/dev/null 2>&1 || true
REMOTECMD
)
if sudo -u "$ACTUAL_USER" ssh -t "${user}@${host}" "$remote_cmd"; then
log_success "Remote share [$share_name] -> $remote_path configured and smbd restarted."
log_success "Remote share [$share_name] -> $remote_path configured, Samba account '$user' set, smbd restarted."
else
log_error "Remote Samba configuration failed — check the output above."
return 1
@@ -218,13 +293,26 @@ REMOTECMD
# ── Local mount + fstab ─────────────────────────────────────────────────────
_vdm_mount_local() {
local host="$1" share_name="$2" mount_point="$3" label="$4"
local host="$1" share_name="$2" mount_point="$3" label="$4" smb_user="$5" smb_pass="$6"
command -v mount.cifs >/dev/null 2>&1 || apt-get install -y cifs-utils -qq
mkdir -p "$mount_point"
local opts="guest,uid=$(id -u "$ACTUAL_USER"),gid=$(id -g "$ACTUAL_USER"),iocharset=utf8,nofail,_netdev"
# Credentials file, not a guest/inline password — root-only, matching
# tools/mount-network-drive.sh's existing convention for SMB creds.
local creds_file="/etc/samba/credentials.vpn-data-mount-${label}"
mkdir -p /etc/samba
cat > "$creds_file" << CREDS
username=${smb_user}
password=${smb_pass}
CREDS
chmod 600 "$creds_file"
chown root:root "$creds_file"
# sec=ntlmssp explicitly — see the file header on errno 79/ENOKEY.
# Modern default and what a credentialed mount should use anyway.
local opts="credentials=${creds_file},sec=ntlmssp,uid=$(id -u "$ACTUAL_USER"),gid=$(id -g "$ACTUAL_USER"),iocharset=utf8,nofail,_netdev"
local share="//${host}/${share_name}"
log_info "Testing mount..."
@@ -266,16 +354,33 @@ _vdm_add_mount() {
break
done
local HOST="" SSH_USER=""
prompt_text " Home box's NetBird IP (or hostname — check 'netbird status' on that box):" "" HOST
if [ -z "$HOST" ]; then
local HOST_INPUT="" HOST="" SSH_USER=""
prompt_text " Home box's NetBird IP or an already-named host (check 'netbird status' on that box):" "" HOST_INPUT
if [ -z "$HOST_INPUT" ]; then
log_warning "No host entered — cancelling this mount."
return 1
fi
_vdm_resolve_host "$HOST_INPUT" "$LABEL"
HOST="$RESOLVED_HOST"
prompt_text " SSH username on the home box:" "$ACTUAL_USER" SSH_USER
_vdm_ensure_ssh_trust "$SSH_USER" "$HOST" || return 1
# Pure convenience on top of the /etc/hosts naming above (which is what
# actually makes the mount itself usable by name) — an SSH Host alias
# additionally skips typing the username for interactive `ssh` use.
# Only offered for a genuinely new name, not every time this mount (or
# another one on the same host) is set up.
if [[ ! "$HOST" =~ ^[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+$ ]] \
&& declare -F add_ssh_host_alias >/dev/null 2>&1 \
&& declare -F ssh_host_alias_exists >/dev/null 2>&1 \
&& ! ssh_host_alias_exists "$HOST"; then
local ADD_ALIAS=""
prompt_yn " Also add '$HOST' as an SSH alias (ssh $HOST, no username needed)? (y/n):" "y" ADD_ALIAS
[[ "$ADD_ALIAS" =~ ^[Yy]$ ]] && add_ssh_host_alias "$HOST" "$HOST_INPUT" "$SSH_USER" "22"
fi
local REMOTE_PATH="" MOUNT_POINT=""
prompt_text " Path on the home box to share (e.g. /home/${SSH_USER}/media):" "" REMOTE_PATH
if [ -z "$REMOTE_PATH" ]; then
@@ -285,7 +390,13 @@ _vdm_add_mount() {
prompt_text " Local mount point:" "/mnt/${LABEL}" MOUNT_POINT
_vdm_setup_remote_samba "$SSH_USER" "$HOST" "$REMOTE_PATH" "$LABEL" || return 1
_vdm_mount_local "$HOST" "$LABEL" "$MOUNT_POINT" "$LABEL" || return 1
_vdm_mount_local "$HOST" "$LABEL" "$MOUNT_POINT" "$LABEL" "$SSH_USER" "$SMB_PASS" || return 1
# Read by callers like services/filebrowser.sh/audiobookshelf.sh/emby.sh
# that chain into this service and want to default their own "which
# directory" prompt to whatever was just mounted, without needing to
# know or guess the path themselves.
VDM_LAST_MOUNT_POINT="$MOUNT_POINT"
echo ""
log_success "Done: $HOST:$REMOTE_PATH is now mounted at $MOUNT_POINT"
@@ -300,9 +411,10 @@ install_vpn-data-mount() {
echo "╚══════════════════════════════════════════════════════════╝"
if [ "$DRY_RUN" = true ]; then
echo "[DRY-RUN] Would offer to name a raw IP in /etc/hosts for reuse (SSH + this mount)"
echo "[DRY-RUN] Would test/set up passwordless SSH to a home box over its NetBird IP"
echo "[DRY-RUN] Would remotely install+configure Samba there for a chosen path"
echo "[DRY-RUN] Would mount it locally over CIFS and add it to /etc/fstab"
echo "[DRY-RUN] Would remotely install Samba + configure a share and a dedicated account there"
echo "[DRY-RUN] Would mount it locally over CIFS (credentials file, not guest) and add it to /etc/fstab"
echo "[DRY-RUN] Repeatable — can be run again for additional home boxes/shares"
return 0
fi
+8 -3
View File
@@ -316,10 +316,15 @@ install_wordpress() {
[ -n "$WP_ADMIN_PASS" ] || WP_ADMIN_PASS="$(generate_password 16)"
# ── Free host port (multiple sites can't all bind the same one) ─────────
# Was checking `docker ps -a` port lists — only catches ports Docker
# itself currently has bound, not ports held by non-Docker processes or
# anything else on the host. Confirmed live: this let a fresh site pick
# an already-occupied port ("address already in use" at container
# start). find_free_port checks actual OS-level listening sockets via
# ss, the same convention every other service here follows — see
# CLAUDE.md's "Port collision avoidance" section.
local WEB_PORT=8090
while docker ps -a --format '{{.Ports}}' 2>/dev/null | grep -q ":${WEB_PORT}->"; do
WEB_PORT=$((WEB_PORT + 1))
done
find_free_port WEB_PORT "$WEB_PORT"
mkdir -p "$DIR/html" "$DIR/db" "$DIR/uploads-ini.d"
ensure_docker_dir_ownership "$DIR"