Services & daemons

The Services section of a server (/servers/:id/services) is where you control the processes that keep the box running. It has two tabs: System services — systemd units like Nginx, PHP-FPM, MariaDB and Redis — and Daemons — long-running application processes (queue workers, custom scripts) kept alive by Supervisor. It’s available on all plans.

Restart a service, step by step

1. Open the System services tab (the default). Each row is a systemd unit with its status, memory, uptime and PID.
2. Hover the row — the status badge swaps to an action group. Click Restart (or Start / Stop depending on the current state).
3. The row shows a spinner while systemctl runs, then flips to the new status. The agent re-pushes its inventory so memory, uptime and PID refresh a moment later.

To dig deeper, click Details (the eye icon) to open a drawer with the process facts, an Auto-start on boot toggle, and a live tail of the unit’s journal.

What else you can do

System services tab

• Filter and search — filter by status (All / Running / Stopped / Failed) and search across service name and description.
• Start / Stop / Restart any managed unit from the row’s hover actions or the details drawer.
• Toggle auto-start on boot — the details drawer switch runs systemctl enable / disable.
• Read the journal — the details drawer tails the last ~60 lines of the unit’s journal, colour-coded by level.
• Refresh — ask the agent to re-scan and re-push the current service state.
• MariaDB opens its own unified manage hub (config + runtime + logs) instead of the generic details drawer.

Daemons tab

• Add daemon — register a long-running command (name, command, run-as user, working directory, process count, autostart/autorestart). Supervisor keeps it running and restarts it on crash.
• Start / Stop / Restart a daemon from its card.
• View logs — tail the daemon’s stdout in a drawer (last ~200 lines).
• Edit — change the command, user, directory or process count. The name is fixed once created (delete and recreate to rename).
• Delete — stops the process and removes its Supervisor config.
• Search across daemon name and command.

How it works

Actions on both tabs dispatch jobs to the server’s agent, which runs them natively:

Action	Job	Agent runs
List services	services inventory	reads the systemd unit state
Start / Stop / Restart / Enable / Disable	service.action	systemctl <action> <unit>
Service journal	service.logs	journalctl -u <unit>
List daemons	daemons inventory	supervisorctl program state
Create / edit daemon	daemon.create	writes a Supervisor program config
Start / Stop / Restart daemon	daemon.action	supervisorctl <action> mz-<name>
Delete daemon	daemon.delete	removes the Supervisor config
Daemon logs	daemon.logs	tails the program’s stdout log

Both lists paint instantly from the control-plane inventory cache (the agent pushes services and daemons snapshots), so the page loads without waiting on a live call. Mutations run live and the agent re-pushes the affected inventory, so the UI self-corrects within a second.

System services are limited to an allowlist of units MZPanel manages — Nginx, MariaDB, redis-server, Docker, cron, SSH, fail2ban, every phpX.Y-fpm, plus extension services (PostgreSQL, MongoDB, Memcached, Supervisor, BIND, Varnish, Netdata, ClamAV). Anything outside that set is neither shown nor actionable, so a stray click can never touch an arbitrary system unit. Each panel-created daemon is a Supervisor program named mz-<name>, kept separate from any Supervisor programs you manage by hand.

CLI & automation

Every action maps to an mz command on the box — the same engine the dashboard drives. SSH in and run them directly, or let an on-box AI (ClaudeCode) run them:

mz service list --json                         # every managed unit + its state
mz service restart nginx --json                # start | stop | restart | enable | disable
mz service logs php8.3-fpm --lines 100 --json  # tail a unit's journal
mz daemon create --name queue-default \
   --command "php artisan queue:work --tries=3" --user www-data --json
mz daemon restart queue-default --json         # start | stop | restart
mz daemon logs queue-default --json            # tail the daemon's stdout

Service names must be on the allowlist above; mz service / mz daemon with no sub-action errors. See The mz CLI and the command catalog.

Gotchas & troubleshooting

• Actions need the agent online. The lists render from the last cached snapshot when the server is offline (you’ll see a preview banner), but start/stop/restart, create and live logs all require the agent to be connected.
• Only allowlisted units appear. If a service you expect is missing, it’s not in MZPanel’s managed set — that’s by design, not a bug. Extension services (PostgreSQL, Mongo, ClamAV…) show up once the extension is installed.
• A stopped service reads ”—” for memory/uptime/PID. Those are live process facts; they only exist while the unit is running.
• A bad daemon command shows up as “failed,” not a save error. Supervisor accepts the config but the process exits immediately (FATAL). MZPanel surfaces Supervisor’s own message inline — check the daemon’s Logs and fix the command or working directory, then Restart.
• Daemons can’t be renamed. The name keys the Supervisor program; change your mind by deleting and recreating.

Related

• Cron jobs — scheduled tasks on the same box.
• OS updates — package updates and reboots for the box.
• The mz CLI — driving the box from the command line.