Background services¶

Besides the webserver, Userli needs two long-running processes:

Process	Command	Purpose
Worker	`bin/console messenger:consume async`	Sends emails (welcome, recovery, notifications, reports) and webhooks, and runs the maintenance tasks.
Scheduler	`bin/console messenger:consume scheduler_maintenance`	Triggers the recurring maintenance tasks on schedule.

Warning

Without the worker, no emails and no webhooks are sent and maintenance tasks never run. Without the scheduler, maintenance tasks only run when started manually in Admin → Maintenance.

Both processes should be restarted regularly to free memory and to pick up new code after updates. The --time-limit option makes them exit after the given number of seconds; the process supervisor starts them again.

systemd¶

Create one unit per process. Adjust user, path and PHP binary.

/etc/systemd/system/userli-worker.service

[Unit]
Description=Userli worker
After=network.target mariadb.service

[Service]
User=userli
Group=userli
WorkingDirectory=/var/www/userli
ExecStart=/usr/bin/php bin/console messenger:consume async --time-limit=3600 --memory-limit=128M
Restart=always
RestartSec=5

[Install]
WantedBy=multi-user.target

/etc/systemd/system/userli-scheduler.service

[Unit]
Description=Userli scheduler
After=network.target mariadb.service

[Service]
User=userli
Group=userli
WorkingDirectory=/var/www/userli
ExecStart=/usr/bin/php bin/console messenger:consume scheduler_maintenance --time-limit=3600 --memory-limit=128M
Restart=always
RestartSec=5

[Install]
WantedBy=multi-user.target

Enable and start them:

sudo systemctl daemon-reload
sudo systemctl enable --now userli-worker userli-scheduler

Queue¶

Queued messages are stored in the database by default. To use Redis or AMQP, set MESSENGER_TRANSPORT_DSN, see Configuration.

Failed messages are retried three times (after 10 seconds, 1 minute and 6 minutes). To inspect the queue:

bin/console messenger:stats

Scheduler state¶

The scheduler remembers when tasks ran last. If it was not running, it catches up missed tasks on start. The state is stored in the cache – with several hosts or containers, use Redis so that all share the same state. Run only one scheduler process.