Writing Service Units

# /etc/systemd/system/myapp.service
[Unit]
Description=My Application Server
Documentation=https://docs.myapp.com
After=network.target postgresql.service
Requires=postgresql.service
Wants=redis.service
StartLimitIntervalSec=300
StartLimitBurst=5

[Service]
Type=simple
User=appuser
Group=appgroup
WorkingDirectory=/opt/myapp
Environment=NODE_ENV=production
EnvironmentFile=/etc/myapp/env
ExecStartPre=/usr/bin/test -f /opt/myapp/config.yaml
ExecStart=/opt/myapp/bin/server --config /opt/myapp/config.yaml
ExecReload=/bin/kill -HUP $MAINPID
ExecStop=/bin/kill -TERM $MAINPID
Restart=on-failure
RestartSec=5
TimeoutStartSec=30
TimeoutStopSec=30
KillMode=mixed
KillSignal=SIGTERM
StandardOutput=journal
StandardError=journal
SyslogIdentifier=myapp

[Install]
WantedBy=multi-user.target

# Type=simple (default)
# Process stays in foreground, systemd considers it started immediately
[Service]
Type=simple
ExecStart=/usr/bin/myapp

# Type=forking
# Process forks, parent exits. systemd tracks the child.
[Service]
Type=forking
PIDFile=/run/myapp.pid
ExecStart=/usr/sbin/myapp --daemon

# Type=oneshot
# Runs once, exits. Used with RemainAfterExit=yes for state tracking.
[Service]
Type=oneshot
RemainAfterExit=yes
ExecStart=/usr/bin/setup-script.sh

# Type=notify
# Process calls sd_notify() when ready. Best for applications that support it.
[Service]
Type=notify
ExecStart=/usr/bin/myapp

# Restart policies
# no        — Never restart (default)
# always    — Always restart
# on-success — Restart only on clean exit (code 0)
# on-failure — Restart on non-zero exit, signal, timeout, or watchdog
# on-abnormal — Restart on signal, timeout, or watchdog (not exit code)
# on-abort   — Restart only on unclean signal
# on-watchdog — Restart only on watchdog timeout

# Common patterns
Restart=on-failure       # Restart on errors
RestartSec=5             # Wait 5 seconds before restart
StartLimitBurst=5        # Max 5 restarts...
StartLimitIntervalSec=300 # ...within 300 seconds

# Rate limiting prevents restart loops
[Service]
Restart=on-failure
RestartSec=5
StartLimitBurst=5
StartLimitIntervalSec=300

# If restart limit is hit, the unit enters failed state
# Check with: systemctl show myapp -p NRestarts

# /etc/myapp/env
NODE_ENV=production
DATABASE_URL=postgresql://localhost/myapp
REDIS_URL=redis://localhost:6379
SECRET_KEY=abc123def456
PORT=8080

# In the service unit
[Service]
EnvironmentFile=/etc/myapp/env
Environment=NODE_ENV=production
EnvironmentFile=-/etc/myapp/env.local  # Optional (ignore if missing)

# Reference in ExecStart
ExecStart=/opt/myapp/bin/server --port $PORT

# Multiple environment files (processed in order)
EnvironmentFile=/etc/myapp/env
EnvironmentFile=/etc/myapp/env.secrets

# Permissions
# chmod 600 /etc/myapp/env
# chown root:appuser /etc/myapp/env

# Create drop-in directory
sudo mkdir -p /etc/systemd/system/myapp.service.d/

# Create override file
cat > /etc/systemd/system/myapp.service.d/override.conf << 'EOF'
[Service]
Environment=DEBUG=true
LimitNOFILE=65536
Nice=5

[Unit]
After=redis.service
EOF

# Reload systemd
sudo systemctl daemon-reload

# View effective configuration
systemctl cat myapp.service

# View all overrides
systemctl show myapp.service -p DropInPaths

# Reset to original
sudo rm /etc/systemd/system/myapp.service.d/override.conf
sudo systemctl daemon-reload

# Verify syntax
systemd-analyze verify /etc/systemd/system/myapp.service

# Check configuration
systemd-analyze cat-config systemd/system/myapp.service

# Test without enabling
systemctl start myapp.service
systemctl status myapp.service
systemctl stop myapp.service

# Analyze dependencies
systemd-analyze dot myapp.service | dot -Tpng > deps.png

# Check for errors
journalctl -u myapp.service -n 50

# Validate all unit files
systemd-analyze verify /etc/systemd/system/*.service