Files
Charon/.docker/docker-entrypoint.sh
GitHub Actions 892b89fc9d feat: break-glass security reset
Implement dual-registry container publishing to both GHCR and Docker Hub
for maximum distribution reach. Add emergency security reset endpoint
("break-glass" mechanism) to recover from ACL lockout situations.

Key changes:

Docker Hub + GHCR dual publishing with Cosign signing and SBOM
Emergency reset endpoint POST /api/v1/emergency/security-reset
Token-based authentication bypasses Cerberus middleware
Rate limited (5/hour) with audit logging
30 new security enforcement E2E tests covering ACL, WAF, CrowdSec,
Rate Limiting, Security Headers, and Combined scenarios
Fixed container startup permission issue (tmpfs directory ownership)
Playwright config updated with testIgnore for browser projects
Security: Token via CHARON_EMERGENCY_TOKEN env var (32+ chars recommended)
Tests: 689 passed, 86% backend coverage, 85% frontend coverage
2026-01-25 20:14:06 +00:00

354 lines
16 KiB
Bash
Executable File

#!/bin/sh
set -e
# Entrypoint script to run both Caddy and Charon in a single container
# This simplifies deployment for home users
echo "Starting Charon with integrated Caddy..."
is_root() {
[ "$(id -u)" -eq 0 ]
}
run_as_charon() {
if is_root; then
gosu charon "$@"
else
"$@"
fi
}
# ============================================================================
# Volume Permission Handling for Non-Root User
# ============================================================================
# When running as non-root user (charon), mounted volumes may have incorrect
# permissions. This section ensures the application can write to required paths.
# Note: This runs as the charon user, so we can only fix owned directories.
# Ensure /app/data exists and is writable (primary data volume)
if [ ! -w "/app/data" ] 2>/dev/null; then
echo "Warning: /app/data is not writable. Please ensure volume permissions are correct."
echo " Run: docker run ... -v charon_data:/app/data ..."
echo " Or fix permissions: chown -R 1000:1000 /path/to/volume"
fi
# Ensure /config exists and is writable (Caddy config volume)
if [ ! -w "/config" ] 2>/dev/null; then
echo "Warning: /config is not writable. Please ensure volume permissions are correct."
fi
# Create required subdirectories in writable volumes
mkdir -p /app/data/caddy 2>/dev/null || true
mkdir -p /app/data/crowdsec 2>/dev/null || true
mkdir -p /app/data/geoip 2>/dev/null || true
# Fix ownership for directories created as root
if is_root; then
chown -R charon:charon /app/data/caddy 2>/dev/null || true
chown -R charon:charon /app/data/crowdsec 2>/dev/null || true
chown -R charon:charon /app/data/geoip 2>/dev/null || true
fi
# ============================================================================
# Plugin Directory Permission Verification
# ============================================================================
# The PluginLoaderService requires the plugin directory to NOT be world-writable
# (mode 0002 bit must not be set). This is a security requirement to prevent
# malicious plugin injection.
PLUGINS_DIR="${CHARON_PLUGINS_DIR:-/app/plugins}"
if [ -d "$PLUGINS_DIR" ]; then
# Check if directory is world-writable (security risk)
# Using find -perm -0002 is more robust than stat regex - handles sticky/setgid bits correctly
if find "$PLUGINS_DIR" -maxdepth 0 -perm -0002 -print -quit 2>/dev/null | grep -q .; then
echo "⚠️ WARNING: Plugin directory $PLUGINS_DIR is world-writable!"
echo " This is a security risk - plugins could be injected by any user."
echo " Attempting to fix permissions (removing world-writable bit)..."
# Use chmod o-w to only remove world-writable, preserving sticky/setgid bits
if chmod o-w "$PLUGINS_DIR" 2>/dev/null; then
echo " ✓ Fixed: Plugin directory world-writable permission removed"
else
echo " ✗ ERROR: Cannot fix permissions. Please run: chmod o-w $PLUGINS_DIR"
echo " Plugin loading may fail due to insecure permissions."
fi
else
echo "✓ Plugin directory permissions OK: $PLUGINS_DIR"
fi
else
echo "Note: Plugin directory $PLUGINS_DIR does not exist (plugins disabled)"
fi
# ============================================================================
# Docker Socket Permission Handling
# ============================================================================
# The Docker integration feature requires access to the Docker socket.
# If the container runs as root, we can auto-align group membership with the
# socket GID. If running non-root (default), we cannot modify groups; users
# can enable Docker integration by using a compatible GID / --group-add.
if [ -S "/var/run/docker.sock" ] && is_root; then
DOCKER_SOCK_GID=$(stat -c '%g' /var/run/docker.sock 2>/dev/null || echo "")
if [ -n "$DOCKER_SOCK_GID" ] && [ "$DOCKER_SOCK_GID" != "0" ]; then
# Check if a group with this GID exists
if ! getent group "$DOCKER_SOCK_GID" >/dev/null 2>&1; then
echo "Docker socket detected (gid=$DOCKER_SOCK_GID) - creating docker group and adding charon user..."
# Create docker group with the socket's GID
groupadd -g "$DOCKER_SOCK_GID" docker 2>/dev/null || true
# Add charon user to the docker group
usermod -aG docker charon 2>/dev/null || true
echo "Docker integration enabled for charon user"
else
# Group exists, just add charon to it
GROUP_NAME=$(getent group "$DOCKER_SOCK_GID" | cut -d: -f1)
echo "Docker socket detected (gid=$DOCKER_SOCK_GID, group=$GROUP_NAME) - adding charon user..."
usermod -aG "$GROUP_NAME" charon 2>/dev/null || true
echo "Docker integration enabled for charon user"
fi
fi
elif [ -S "/var/run/docker.sock" ]; then
echo "Note: Docker socket mounted but container is running non-root; skipping docker.sock group setup."
echo " If Docker discovery is needed, run with matching group permissions (e.g., --group-add)"
else
echo "Note: Docker socket not found. Docker container discovery will be unavailable."
fi
# ============================================================================
# CrowdSec Initialization
# ============================================================================
# Note: CrowdSec agent is not auto-started. Lifecycle is GUI-controlled via backend handlers.
# Initialize CrowdSec configuration if cscli is present
if command -v cscli >/dev/null; then
echo "Initializing CrowdSec configuration..."
# Define persistent paths
CS_PERSIST_DIR="/app/data/crowdsec"
CS_CONFIG_DIR="$CS_PERSIST_DIR/config"
CS_DATA_DIR="$CS_PERSIST_DIR/data"
CS_LOG_DIR="/var/log/crowdsec"
# Ensure persistent directories exist (within writable volume)
mkdir -p "$CS_CONFIG_DIR" 2>/dev/null || echo "Warning: Cannot create $CS_CONFIG_DIR"
mkdir -p "$CS_DATA_DIR" 2>/dev/null || echo "Warning: Cannot create $CS_DATA_DIR"
mkdir -p "$CS_PERSIST_DIR/hub_cache"
# Log directories are created at build time with correct ownership
# Only attempt to create if they don't exist (first run scenarios)
mkdir -p /var/log/crowdsec 2>/dev/null || true
mkdir -p /var/log/caddy 2>/dev/null || true
# Initialize persistent config if key files are missing
if [ ! -f "$CS_CONFIG_DIR/config.yaml" ]; then
echo "Initializing persistent CrowdSec configuration..."
if [ -d "/etc/crowdsec.dist" ] && [ -n "$(ls -A /etc/crowdsec.dist 2>/dev/null)" ]; then
cp -r /etc/crowdsec.dist/* "$CS_CONFIG_DIR/" || {
echo "ERROR: Failed to copy config from /etc/crowdsec.dist"
exit 1
}
echo "Successfully initialized config from .dist directory"
elif [ -d "/etc/crowdsec" ] && [ ! -L "/etc/crowdsec" ] && [ -n "$(ls -A /etc/crowdsec 2>/dev/null)" ]; then
cp -r /etc/crowdsec/* "$CS_CONFIG_DIR/" || {
echo "ERROR: Failed to copy config from /etc/crowdsec"
exit 1
}
echo "Successfully initialized config from /etc/crowdsec"
else
echo "ERROR: No config source found (neither .dist nor /etc/crowdsec available)"
exit 1
fi
fi
# Verify symlink exists (created at build time)
# Note: Symlink is created in Dockerfile as root before switching to non-root user
# Non-root users cannot create symlinks in /etc, so this must be done at build time
if [ -L "/etc/crowdsec" ]; then
echo "CrowdSec config symlink verified: /etc/crowdsec -> $CS_CONFIG_DIR"
else
echo "WARNING: /etc/crowdsec symlink not found. This may indicate a build issue."
echo "Expected: /etc/crowdsec -> /app/data/crowdsec/config"
# Try to continue anyway - config may still work if CrowdSec uses CFG env var
fi
# Create/update acquisition config for Caddy logs
if [ ! -f "/etc/crowdsec/acquis.yaml" ] || [ ! -s "/etc/crowdsec/acquis.yaml" ]; then
echo "Creating acquisition configuration for Caddy logs..."
cat > /etc/crowdsec/acquis.yaml << 'ACQUIS_EOF'
# Caddy access logs acquisition
# CrowdSec will monitor these files for security events
source: file
filenames:
- /var/log/caddy/access.log
- /var/log/caddy/*.log
labels:
type: caddy
ACQUIS_EOF
fi
# Ensure hub directory exists in persistent storage
mkdir -p /etc/crowdsec/hub
# Perform variable substitution
export CFG=/etc/crowdsec
export DATA="$CS_DATA_DIR"
export PID=/var/run/crowdsec.pid
export LOG="$CS_LOG_DIR/crowdsec.log"
# Process config.yaml and user.yaml with envsubst
# We use a temp file to avoid issues with reading/writing same file
for file in /etc/crowdsec/config.yaml /etc/crowdsec/user.yaml; do
if [ -f "$file" ]; then
envsubst < "$file" > "$file.tmp" && mv "$file.tmp" "$file"
chown charon:charon "$file" 2>/dev/null || true
fi
done
# Configure CrowdSec LAPI to use port 8085 to avoid conflict with Charon (port 8080)
if [ -f "/etc/crowdsec/config.yaml" ]; then
sed -i 's|listen_uri: 127.0.0.1:8080|listen_uri: 127.0.0.1:8085|g' /etc/crowdsec/config.yaml
sed -i 's|listen_uri: 0.0.0.0:8080|listen_uri: 127.0.0.1:8085|g' /etc/crowdsec/config.yaml
fi
# Update local_api_credentials.yaml to use correct port
if [ -f "/etc/crowdsec/local_api_credentials.yaml" ]; then
sed -i 's|url: http://127.0.0.1:8080|url: http://127.0.0.1:8085|g' /etc/crowdsec/local_api_credentials.yaml
sed -i 's|url: http://localhost:8080|url: http://127.0.0.1:8085|g' /etc/crowdsec/local_api_credentials.yaml
fi
# Fix log directory path (ensure it points to /var/log/crowdsec/ not /var/log/)
sed -i 's|log_dir: /var/log/$|log_dir: /var/log/crowdsec/|g' "$CS_CONFIG_DIR/config.yaml"
# Also handle case where it might be without trailing slash
sed -i 's|log_dir: /var/log$|log_dir: /var/log/crowdsec|g' "$CS_CONFIG_DIR/config.yaml"
# Verify LAPI configuration was applied correctly
if grep -q "listen_uri:.*:8085" "$CS_CONFIG_DIR/config.yaml"; then
echo "✓ CrowdSec LAPI configured for port 8085"
else
echo "✗ WARNING: LAPI port configuration may be incorrect"
fi
# Update hub index to ensure CrowdSec can start
if [ ! -f "/etc/crowdsec/hub/.index.json" ]; then
echo "Updating CrowdSec hub index..."
timeout 60s cscli hub update 2>/dev/null || echo "⚠️ Hub update timed out or failed, continuing..."
fi
# Ensure local machine is registered (auto-heal for volume/config mismatch)
# We force registration because we just restored configuration (and likely credentials)
echo "Registering local machine..."
cscli machines add -a --force 2>/dev/null || echo "Warning: Machine registration may have failed"
# Install hub items (parsers, scenarios, collections) if local mode enabled
if [ "$SECURITY_CROWDSEC_MODE" = "local" ]; then
echo "Installing CrowdSec hub items..."
if [ -x /usr/local/bin/install_hub_items.sh ]; then
/usr/local/bin/install_hub_items.sh 2>/dev/null || echo "Warning: Some hub items may not have installed"
fi
fi
# Fix ownership AFTER cscli commands (they run as root and create root-owned files)
echo "Fixing CrowdSec file ownership..."
if is_root; then
chown -R charon:charon /var/lib/crowdsec 2>/dev/null || true
chown -R charon:charon /app/data/crowdsec 2>/dev/null || true
chown -R charon:charon /var/log/crowdsec 2>/dev/null || true
fi
fi
# CrowdSec Lifecycle Management:
# CrowdSec configuration is initialized above (symlinks, directories, hub updates)
# However, the CrowdSec agent is NOT auto-started in the entrypoint.
# Instead, CrowdSec lifecycle is managed by the backend handlers via GUI controls.
# This makes CrowdSec consistent with other security features (WAF, ACL, Rate Limiting).
# Users enable/disable CrowdSec using the Security dashboard toggle, which calls:
# - POST /api/v1/admin/crowdsec/start (to start the agent)
# - POST /api/v1/admin/crowdsec/stop (to stop the agent)
# This approach provides:
# - Consistent user experience across all security features
# - No environment variable dependency
# - Real-time control without container restart
# - Proper integration with Charon's security orchestration
echo "CrowdSec configuration initialized. Agent lifecycle is GUI-controlled."
# Start Caddy in the background with initial empty config
# Run Caddy as charon user for security
echo '{"admin":{"listen":"0.0.0.0:2019"},"apps":{}}' > /config/caddy.json
# Use JSON config directly; no adapter needed
run_as_charon caddy run --config /config/caddy.json &
CADDY_PID=$!
echo "Caddy started (PID: $CADDY_PID)"
# Wait for Caddy to be ready
echo "Waiting for Caddy admin API..."
i=1
while [ "$i" -le 30 ]; do
if curl -sf http://127.0.0.1:2019/config/ > /dev/null 2>&1; then
echo "Caddy is ready!"
break
fi
i=$((i+1))
sleep 1
done
# Start Charon management application
# Drop privileges to charon user before starting the application
# This maintains security while allowing Docker socket access via group membership
# Note: When running as root, we use gosu; otherwise we run directly.
echo "Starting Charon management application..."
DEBUG_FLAG=${CHARON_DEBUG:-$CPMP_DEBUG}
DEBUG_PORT=${CHARON_DEBUG_PORT:-${CPMP_DEBUG_PORT:-2345}}
# Determine binary path
bin_path=/app/charon
if [ ! -f "$bin_path" ]; then
bin_path=/app/cpmp
fi
if [ "$DEBUG_FLAG" = "1" ]; then
# Check if binary has debug symbols (required for Delve)
# objdump -h lists section headers; .debug_info is present if DWARF symbols exist
if command -v objdump >/dev/null 2>&1; then
if ! objdump -h "$bin_path" 2>/dev/null | grep -q '\.debug_info'; then
echo "⚠️ WARNING: Binary lacks debug symbols (DWARF info stripped)."
echo " Delve debugging will NOT work with this binary."
echo " To fix, rebuild with: docker build --build-arg BUILD_DEBUG=1 ..."
echo " Falling back to normal execution (without debugger)."
run_as_charon "$bin_path" &
else
echo "✓ Debug symbols detected. Running Charon under Delve (port $DEBUG_PORT)"
run_as_charon /usr/local/bin/dlv exec "$bin_path" --headless --listen=":$DEBUG_PORT" --api-version=2 --accept-multiclient --continue --log -- &
fi
else
# objdump not available, try to run Delve anyway with a warning
echo "Note: Cannot verify debug symbols (objdump not found). Attempting Delve..."
run_as_charon /usr/local/bin/dlv exec "$bin_path" --headless --listen=":$DEBUG_PORT" --api-version=2 --accept-multiclient --continue --log -- &
fi
else
run_as_charon "$bin_path" &
fi
APP_PID=$!
echo "Charon started (PID: $APP_PID)"
shutdown() {
echo "Shutting down..."
kill -TERM "$APP_PID" 2>/dev/null || true
kill -TERM "$CADDY_PID" 2>/dev/null || true
# Note: CrowdSec process lifecycle is managed by backend handlers
# The backend will handle graceful CrowdSec shutdown when the container stops
wait "$APP_PID" 2>/dev/null || true
wait "$CADDY_PID" 2>/dev/null || true
exit 0
}
# Trap signals for graceful shutdown
trap 'shutdown' TERM INT
echo "Charon is running!"
echo " - Management UI: http://localhost:8080"
echo " - Caddy Proxy: http://localhost:80, https://localhost:443"
echo " - Caddy Admin API: http://localhost:2019"
# Wait loop: exit when either process dies, then shutdown the other
while kill -0 "$APP_PID" 2>/dev/null && kill -0 "$CADDY_PID" 2>/dev/null; do
sleep 1
done
echo "A process exited, initiating shutdown..."
shutdown