Install

This page is for the developer or IT person who installs and runs Rangefront Waste. It covers requirements, every setting in .env.example, the queue worker, the scheduler, storage and mail. For shipping to a server, see deploy.

Requirements

Need	Notes
PHP 8.3 or newer	Laravel 13. Needs the usual Laravel extensions plus a database driver (pdo_pgsql or pdo_mysql) and sodium (used to verify Telnyx reply signatures). The demo runs PHP 8.4
Composer	For apps/backend-laravel
Database	PostgreSQL (the dev stack uses the PostGIS image) or MySQL 8. Tests run on SQLite
Redis	Queues, cache and Horizon. The predis client is bundled. Use REDIS_CLIENT=predis if phpredis isn't installed
S3-compatible storage	AWS S3, MinIO or similar, for photos, signatures and PDFs
Headless Chromium	Renders PDFs in production. Set PDF_CHROMIUM_BINARY
A process manager	systemd or Supervisor, to keep Horizon and Reverb running
Cron	To run the scheduler every minute
Node 20.19 or newer, on a build machine	Builds the console and the field app (Vite 7). The server does not need Node

For local development, docker compose -f infra/local-dev/docker-compose.yml up -d starts Postgres with PostGIS (port 54329), Redis (port 63799), SeaweedFS as S3 (port 9000, bucket rangefront-waste, keys rangefront / rangefront-secret, created by the s3-bucket one-shot) and Mailpit (SMTP 1025, web 8025). Postgres 18 stores its cluster under /var/lib/postgresql, so the volume mounts there. The Postgres image is amd64-only and runs under emulation on Apple Silicon. If another local service already holds 127.0.0.1:9000 (a PHP-FPM, for example), use AWS_ENDPOINT=http://localhost:9000, which Docker still serves.

Install steps

cd apps/backend-laravel
composer install --no-dev --optimize-autoloader     # drop --no-dev for development
cp .env.example .env
php artisan key:generate
# edit .env (see below)
php artisan migrate --force
php artisan db:seed --class=CompliancePackSeeder    # state and territory rule packs

Keep the generated APP_KEY safe. Credentials stored for payments, SMS, email and integrations are encrypted with it, and can't be recovered without it.

Create the first business

php artisan rangefront:tenant:create "Acme Pumping" acme-pumping owner@acme.example --timezone=Australia/Brisbane

This creates the business, a Main depot, the four default roles (Owner, Dispatcher, Service Agent, Accounts, with the Owner holding every permission), the commissioning certificate form and the standard customer messages (rules off), then emails the owner an invitation. The owner opens the link, chooses a password and signs in. If mail isn't set up yet the command prints the invitation link so you can pass it on. Set CONSOLE_URL (or APP_URL) first, because the link is built from it.

Other commands:

Command	Does
rangefront:tenant:list	Lists businesses with status and how many people can sign in
rangefront:tenant:suspend <org> [--reason=]	Switches a business off. Its people get a clear "business is suspended" message in the console and the field app; nothing is deleted
rangefront:tenant:resume <org>	Switches it back on

<org> is the id or the slug. Each is written to the audit log. rangefront:install-defaults <org> [--enable-messages] is still there for a business made another way, and is safe to run again.

php artisan db:seed creates two demo businesses with fake data, for demos and development only.

Build the clients

npm install
npm run web:build                      # apps/web-aurelia/dist

Serve dist as the site root, with every unknown path falling back to index.html, and send /api, /broadcasting, /sanctum, /storage, /up and /book to Laravel's public/index.php. The websocket path /app/ goes to Reverb. Web build variables:

Variable	Meaning
VITE_RANGEFRONT_API_BASE_URL	API base. Defaults to /api/v1/, which suits a single-host setup
VITE_MAP_TILE_URL	Map tile address. Defaults to OpenStreetMap
VITE_MAP_TILE_ATTRIBUTION	Attribution text for the tiles
VITE_FIELD_APP_URL	Adds a "Field app" link to the console
VITE_FORCE_DEMO_LOGIN, VITE_ENABLE_DEMO_LOGIN, VITE_DEMO_EMAIL, VITE_DEMO_PASSWORD	Auto sign-in with a fixed account, for the public demo only. Without it the console shows its sign-in page. npm run web also auto signs in; set VITE_ENABLE_DEMO_LOGIN=false to see the sign-in page in development. Never bake a real password into a public build

The field app is built with Expo. Its public variables are EXPO_PUBLIC_RANGEFRONT_API_URL (the server email sign-in goes to; leave it empty for a build that only links with codes) and, for demos, EXPO_PUBLIC_RANGEFRONT_DEMO_*. See apps/mobile-expo/AGENTS.md.

Phones on your own server don't need their own build. The store app links to any server with a code from Admin > Work phones. The code's QR holds your APP_URL plus /api/v1, so APP_URL must be the public https address phones reach. A release build refuses a server that isn't https.

Run the processes

Process	Command	Keep running with
Web and API	php-fpm behind nginx, or php artisan serve for development	nginx and php-fpm
Queue worker	php artisan horizon	systemd or Supervisor
Websockets	php artisan reverb:start --host=127.0.0.1 --port=8080	systemd or Supervisor
Scheduler	* * * * * cd /path/to/apps/backend-laravel && php artisan schedule:run >> /dev/null 2>&1	cron

Horizon starts three supervisors that between them run every queue (see Queues). A plain queue:work is not enough, because it only listens to the queues you name.

Restart Horizon after every deploy (php artisan horizon:terminate, or restart the unit) so workers load the new code.

The Horizon dashboard (/horizon) is closed to everyone outside local development. To open it, put the emails of your platform administrators (people who run the installation, not a customer's staff) in PLATFORM_ADMIN_EMAILS, comma separated, and sign in to the Horizon page itself with one of those accounts' console email and password (the browser shows a basic-auth prompt; use HTTPS). Leave it empty to keep Horizon closed. The setting is read from config/access.php.

Check the install with GET /up (see Health check) and php artisan about.

Environment variables

Everything below is in apps/backend-laravel/.env.example unless the table says otherwise.

App

Variable	Meaning
APP_NAME	Product name. Also the default mail sender name
APP_ENV	local, production and so on. production turns the PDF renderer to headless Chromium and strict plugin catalogue checks off
APP_KEY	Encryption key. Generate with key:generate and back it up
APP_DEBUG	false in production
APP_URL	The public address of the site. Pay, quote, tracking and portal links, reply webhooks and OAuth callbacks are built from it
CONSOLE_URL	Where the console is served, when that isn't APP_URL. Invitation and password reset links point here
PLATFORM_ADMIN_EMAILS	Comma separated emails allowed into Horizon outside local development
INVITE_TTL_DAYS, LOGIN_ATTEMPTS	Days an invitation lasts (7), and wrong passwords allowed per email and address before sign-in locks for a minute (5)
CONSOLE_SESSION_HOURS, CONSOLE_REMEMBER_DAYS	How long a console sign-in lasts (12 hours), or with "Keep me signed in" (30 days)
APP_LOCALE, APP_FALLBACK_LOCALE, APP_FAKER_LOCALE	Language settings
APP_MAINTENANCE_DRIVER	file or cache for maintenance mode
APP_MAINTENANCE_STORE	Cache store for maintenance mode (commented out)
PHP_CLI_SERVER_WORKERS	Workers for php artisan serve (commented out)
BCRYPT_ROUNDS	Password hashing cost

Logging

Variable	Meaning
LOG_CHANNEL	Where logs go. stack
LOG_STACK	Channels in the stack. single
LOG_DEPRECATIONS_CHANNEL	Channel for deprecation notices. null
LOG_LEVEL	Minimum level. Use warning or error in production

Database

Variable	Meaning
DB_CONNECTION	pgsql, mysql or sqlite
DB_HOST, DB_PORT	Database server. The dev stack uses 127.0.0.1 and 54329
DB_DATABASE, DB_USERNAME, DB_PASSWORD	Database and login

On MySQL, string columns in composite indexes are limited, so AppServiceProvider sets the default string length to 191 when the connection is mysql.

Session, cache and queue

Variable	Meaning
SESSION_DRIVER	database. The API uses bearer tokens, so sessions matter little. The console's own sign-in lasts CONSOLE_SESSION_HOURS (12) or, with "Keep me signed in", CONSOLE_REMEMBER_DAYS (30)
SESSION_LIFETIME	Minutes
SESSION_ENCRYPT	Encrypt session data
SESSION_PATH, SESSION_DOMAIN	Cookie scope
BROADCAST_CONNECTION	reverb for live updates, or null to turn them off
FILESYSTEM_DISK	Default storage disk. s3
QUEUE_CONNECTION	redis in production. sync runs jobs inline
DB_QUEUE_TABLE	Table name if the queue uses the database (queue_jobs)
CACHE_STORE	redis
CACHE_PREFIX	Key prefix. Set it when several apps share Redis (commented out)
REDIS_CLIENT	phpredis or predis
REDIS_HOST, REDIS_PASSWORD, REDIS_PORT	Redis server. .env.example is set for the dev Docker stack, which publishes Redis on 63799. On a server, set the port of your own Redis, usually 6379
REDIS_QUEUE_RETRY_AFTER	Seconds before a job a worker has not finished is handed to another worker. Default 600. Keep it above the longest Horizon timeout (300) or jobs run twice

When sharing a Redis server, use dedicated REDIS_DB and REDIS_CACHE_DB numbers so apps don't collide, and set HORIZON_PREFIX.

Mail

The built-in email service uses these. Each business can switch to its own provider in Admin, which ignores them. See SMS and email setup.

Variable	Meaning
MAIL_MAILER	log writes mail to the log, smtp sends it
MAIL_SCHEME	smtps or null
MAIL_HOST, MAIL_PORT	SMTP server. .env.example is set for the dev Docker stack, where Mailpit listens on 1025. On a server, use your provider's host and port
MAIL_USERNAME, MAIL_PASSWORD	SMTP login
MAIL_FROM_ADDRESS, MAIL_FROM_NAME	Default sender

Twilio (built-in SMS and the phone agent)

Variable	Meaning
TWILIO_ACCOUNT_SID, TWILIO_AUTH_TOKEN	The server's Twilio account
TWILIO_FROM	Sending number
TWILIO_MESSAGING_SERVICE_SID	A messaging service instead of a number
TWILIO_BASE_URI	API address
TWILIO_STATUS_CALLBACK_URL	Address Twilio posts delivery receipts to, for alert texts
TWILIO_TIMEOUT	Request timeout in seconds
TWILIO_RETRY_AFTER_MINUTES	Wait before retrying a failed alert text
TWILIO_RETRYABLE_ERROR_CODES	Twilio error codes that are worth retrying

Push notifications (Expo)

Variable	Meaning
EXPO_PUSH_URL, EXPO_RECEIPTS_URL	Expo push API addresses
EXPO_ACCESS_TOKEN	Expo access token, if you've enabled push security
EXPO_PUSH_TIMEOUT	Seconds
EXPO_RECEIPT_DELAY_MINUTES	How long to wait before asking for receipts
EXPO_RECEIPT_BATCH_SIZE	Receipts per request
EXPO_RETRY_AFTER_MINUTES	Wait before retrying a failed push

Monitoring alerts

Variable	Meaning
MONITORING_ALERT_RESEND_BATCH_SIZE	Alert notifications resent per run
MONITORING_ALERT_RESEND_MAX_ATTEMPTS	Attempts before giving up

Stripe

Variable	Meaning
STRIPE_SECRET_KEY	Platform secret key. Needed for card payments and for subscription billing
STRIPE_WEBHOOK_SECRET	Signing secret for the subscription billing webhook
STRIPE_API_BASE, STRIPE_API_VERSION, STRIPE_TIMEOUT	API address, pinned version, timeout
STRIPE_WEBHOOK_TOLERANCE_SECONDS	Oldest signature accepted. 300
STRIPE_CHECKOUT_SUCCESS_URL, STRIPE_CHECKOUT_CANCEL_URL, STRIPE_PORTAL_RETURN_URL	Where subscription billing sends the browser back
STRIPE_CONNECT_WEBHOOK_SECRET	Signing secret for the Connect webhook (card payments)
STRIPE_CONNECT_APPLICATION_FEE_PERCENT	Platform fee on card payments. 0 for none
STRIPE_CONNECT_COUNTRY	Country for new connected accounts. AU

See payments setup.

Uploads and antivirus

Variable	Meaning
ATTACHMENT_ANTIVIRUS_DRIVER	basic (built-in checks for unsafe SVG and the EICAR test file) or clamav
ATTACHMENT_ANTIVIRUS_FAIL_OPEN	false rejects a file when the scanner is down. Keep it false
CLAMAV_HOST, CLAMAV_PORT, CLAMAV_UNIX_SOCKET	Where ClamAV listens
CLAMAV_TIMEOUT, CLAMAV_CHUNK_SIZE, CLAMAV_MAX_BYTES	Scan timeout, chunk size and largest file

Council reporting

Variable	Meaning
COUNCIL_REPORT_DISPATCH_ON_SUBMIT	true sends a queued submission straight away. false waits for the 15 minute retry command
COUNCIL_REPORT_DISPATCH_DELAY_SECONDS	Delay before an immediate dispatch
COUNCIL_REPORT_DELIVERY_BATCH_SIZE, COUNCIL_REPORT_RECEIPT_BATCH_SIZE	Submissions handled per run
COUNCIL_REPORT_MAX_ATTEMPTS	Delivery attempts before failing
COUNCIL_REPORT_RETRY_AFTER_MINUTES	Wait between attempts
COUNCIL_REPORT_API_TIMEOUT	Seconds for council API calls
COUNCIL_REPORT_SFTP_DISK	Filesystem disk used for SFTP delivery

Accounting and OAuth

Variable	Meaning
ACCOUNTING_SYNC_INVOICE_BATCH_SIZE	Invoices per sync run
ACCOUNTING_SYNC_API_TIMEOUT	Seconds
XERO_OAUTH_AUTHORIZE_URL, XERO_OAUTH_TOKEN_URL, XERO_OAUTH_CONNECTIONS_URL	Xero endpoints
XERO_OAUTH_REDIRECT_URI	Must match the redirect in your Xero app. APP_URL/api/v1/integrations/oauth/xero/callback
XERO_OAUTH_SCOPES	Space separated scopes
XERO_OAUTH_STATE_TTL_MINUTES	How long a sign-in attempt stays valid
XERO_OAUTH_TIMEOUT	Seconds
XERO_OAUTH_TOKEN_REFRESH_WINDOW_SECONDS	Refresh a token this long before it expires

Route providers

All optional. Without keys the built-in straight-line optimiser is used.

Variable	Meaning
ROUTE_OPTIMIZATION_TRAFFIC_AWARE	Use live traffic where the provider supports it
GOOGLE_ROUTES_API_KEY, GOOGLE_ROUTES_BASE_URI	Google Routes
GOOGLE_ROUTES_TIMEOUT, _RETRIES, _RETRY_SLEEP_MS	Timeout and retry behaviour
GOOGLE_ROUTES_MAX_ELEMENTS	Largest matrix
GOOGLE_ROUTES_TRAFFIC_AWARE	Traffic for Google only
MAPBOX_ACCESS_TOKEN, MAPBOX_BASE_URI	Mapbox
MAPBOX_PROFILE, MAPBOX_TRAFFIC_PROFILE	Routing profiles
MAPBOX_TIMEOUT, MAPBOX_RETRIES, MAPBOX_RETRY_SLEEP_MS	Timeout and retries
MAPBOX_MAX_COORDINATES, MAPBOX_TRAFFIC_MAX_COORDINATES	Stop limits
MAPBOX_TRAFFIC_AWARE	Traffic for Mapbox only
OPENROUTESERVICE_API_KEY, OPENROUTESERVICE_BASE_URI	openrouteservice
OPENROUTESERVICE_PROFILE	Profile, driving-car
OPENROUTESERVICE_TIMEOUT, _RETRIES, _RETRY_SLEEP_MS	Timeout and retries
OPENROUTESERVICE_MAX_LOCATIONS	Stop limit

Object storage

Variable	Meaning
AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY	Storage credentials
AWS_DEFAULT_REGION	Region
AWS_BUCKET	Bucket name
AWS_USE_PATH_STYLE_ENDPOINT	true for MinIO or SeaweedFS
AWS_ENDPOINT	Custom endpoint for MinIO or another S3-compatible service

Realtime (Reverb)

Variable	Meaning
REVERB_APP_ID, REVERB_APP_KEY, REVERB_APP_SECRET	Credentials shared between Laravel and clients
REVERB_HOST, REVERB_PORT, REVERB_SCHEME	The address browsers connect to. Use your public host, port 443 and https in production
REVERB_SERVER_HOST, REVERB_SERVER_PORT	Where the Reverb process listens

Web

Variable	Meaning

Scheduled work and failure alerts

All optional. They are in .env.example, commented out.

Variable	Meaning	Default
OPERATIONS_DUE_JOB_LOOKAHEAD_DAYS	How far ahead draft jobs are created for work falling due	14
OPERATIONS_MOBILE_SYNC_RETENTION_DAYS	Applied field-app sync events older than this are deleted	180
OPERATIONS_LOCATION_EVENT_RETENTION_DAYS	Technician position history older than this is deleted	90
OPERATIONS_UPLOAD_SESSION_RETENTION_DAYS	Uploads never finished, or refused, are cleaned up after this	7
OPERATIONS_COUNCIL_DEADLINE_WARN_HOURS	The office is warned this long before a council deadline	48
OPERATIONS_ROUTE_MATRIX_TTL_DAYS	Cached route legs that upcoming jobs still need are kept this long	30
OPERATIONS_MESSAGE_FROM_HOUR, OPERATIONS_MESSAGE_UNTIL_HOUR	Customer messages only go out between these local hours	8, 20
OPERATIONS_DUE_JOBS_LOCAL_HOUR	The local hour each business gets its due-job run	5
OPERATIONS_CHUNK_SIZE	Rows handled per database chunk	200
HEALTH_SCHEDULER_MAX_AGE_SECONDS	Age of the scheduler heartbeat that fails the health check	300
HEALTH_QUEUE_MAX_BACKLOG	Jobs waiting before the health check fails	1000
FAILED_JOB_ALERT_EMAILS	Platform admins emailed when a job fails for good, comma separated	none
WEBHOOK_ALLOW_HTTP, WEBHOOK_ALLOW_PRIVATE_HOSTS	Let outgoing webhooks use plain http or private and internal addresses. Leave off in production; for local testing only	false
FAILED_JOB_ALERT_THROTTLE_MINUTES	At most one email per job type in this time	15

The first five can be set per business under settings.operations on the organization record, which wins over these values. A business sets the first, how far ahead due work is raised, under Admin > Recurring work. The business timezone decides every "local" hour.

Other settings

These are read by the code. They are in .env.example, commented out.

Variable	Meaning	Default
ANTHROPIC_API_KEY and the other ANTHROPIC_*, AI_*	AI helpers. See AI setup	unset
TWILIO_VOICE, TWILIO_VOICE_WEBHOOK_BASE_URL, PHONE_AGENT_*	The phone agent. See AI setup	see page
TURNSTILE_SITE_KEY, TURNSTILE_SECRET_KEY, TURNSTILE_VERIFY_URL	Server-wide Cloudflare Turnstile for the booking form	unset
PDF_RENDERER	headless_chromium or local	headless_chromium in production
PDF_CHROMIUM_BINARY	Path to Chromium or Chrome	empty
PDF_RENDER_TIMEOUT_SECONDS, PDF_RENDER_MAX_ATTEMPTS	Per-render limits	60, 3
PDF_TEMPLATE_VERSION	Recorded on each PDF	built in
RANGEFRONT_UPLOAD_DISK	Disk for direct uploads	FILESYSTEM_DISK
RANGEFRONT_UPLOAD_URL_TTL	Seconds a signed upload address lasts	900
QUICKBOOKS_OAUTH_*, MYOB_OAUTH_*, SERVICEM8_OAUTH_*, GOOGLE_CALENDAR_OAUTH_*, MICROSOFT_365_OAUTH_*	Authorise and token addresses, REDIRECT_URI and SCOPES for each provider. Set *_REDIRECT_URI to APP_URL/api/v1/integrations/oauth/<provider>/callback	provider defaults
INTEGRATION_OAUTH_STATE_TTL_MINUTES, _TIMEOUT, _TOKEN_REFRESH_WINDOW_SECONDS	Shared OAuth limits	10, 10, 300
EXTENSIBILITY_RETHROW	true lets exceptions from plugins through, for tests	false
EXTENSIBILITY_STRICT	Refuse hooks that aren't in the catalogue	on outside production
HORIZON_PATH, HORIZON_DOMAIN, HORIZON_NAME, HORIZON_PREFIX	The Horizon dashboard address and name, and the Redis key prefix	horizon
DEMO_READ_ONLY	Block all writes. Public demo only	false
SANCTUM_STATEFUL_DOMAINS, SANCTUM_TOKEN_PREFIX	Sanctum. The console does not use Sanctum's stateful cookie mode, so SANCTUM_STATEFUL_DOMAINS can stay at its default	defaults
CONSOLE_COOKIE_NAME, CONSOLE_COOKIE_SAME_SITE, CONSOLE_COOKIE_DOMAIN, CONSOLE_COOKIE_SECURE	The console's httpOnly sign-in cookie. Defaults suit a console and API on one host over HTTPS (Secure is set automatically when the request is HTTPS). Use same_site=none only when they are on different sites	rangefront_console, lax, none, auto
CORS_ALLOWED_ORIGINS	Extra browser origins allowed to call the API with credentials, comma separated. The console origin (CONSOLE_URL or APP_URL) is always allowed. Leave empty when everything is served from one host	empty

Laravel's own variables (DB_URL, REDIS_URL, SESSION_*, LOG_*, AWS_URL and so on) also work as the framework documents.

Queues

Jobs run through Redis, managed by Horizon. config/horizon.php starts three supervisors, in every environment, so each queue has a worker:

Supervisor	Queues	Timeout	Used for
realtime	critical, notifications, webhooks, default	60 s	Customer messages, monitoring alerts, team push, outgoing webhooks, anything unnamed
documents	documents, reports, uploads	180 s	PDF rendering with Chromium, upload scanning
integrations	integrations, council, routing, imports	300 s	Accounting and calendar syncs, council delivery, data imports

Each retries a failed job three times with a growing delay (the supervisor's backoff). Production runs up to 8, 4 and 3 worker processes. Staging and any other environment name (including local) run fewer, so no queue is ever left without a worker. Horizon raises an alert in its dashboard when a queue waits too long (waits in the config).

If you add a queue, add it to a supervisor and to waits. A test (HorizonQueueCoverageTest) reads every queue name in the code and fails if one has no worker, and fails if a timeout is not below REDIS_QUEUE_RETRY_AFTER.

Failed jobs are stored in the database (failed_jobs) for 30 days. php artisan queue:failed lists them and php artisan queue:retry all retries. Set FAILED_JOB_ALERT_EMAILS to have the platform admins emailed when one fails for good, at most once per job type every 15 minutes.

PDF rendering with PDF_RENDERER=local is for development. In production use headless_chromium and make sure the worker's user can run the binary.

Scheduled commands

php artisan schedule:run needs to run every minute. These commands are scheduled in routes/console.php:

Command	When	Does
rangefront:generate-due-jobs	Hourly, acts at 5am local per business	Creates draft jobs from active recurring plans and from systems whose service or pump-out is coming due, then tells dispatch. Also tells dispatch how many booked stops from earlier days were never started
rangefront:send-service-reminders	Hourly	Sends service reminders for upcoming visits, for businesses with a reminder rule on
rangefront:quotes:follow-up	Hourly	Reminds customers about unanswered quotes and expires old ones
rangefront:send-invoice-reminders	Hourly	Sends the due invoice reminders
rangefront:send-statements	Hourly	Sends monthly statements on each business's chosen day
rangefront:council-report-deadline-check	Hourly	Notifies the office when a council report is near, or past, its deadline
rangefront:retry-council-submissions	Every 15 minutes	Delivers queued council submissions and retries failures
rangefront:council-submissions:poll-receipts	Hourly	Checks council receipts
rangefront:expire-upload-sessions	Hourly	Deletes the stored file of uploads abandoned or refused for over 7 days, and marks them expired
rangefront:prune-mobile-sync-events	Daily at 03:30	Deletes applied field-app sync events older than 180 days
rangefront:prune-location-events	Daily at 03:45	Deletes technician position history older than 90 days
rangefront:prune-twig-cache	Weekly, Sunday 04:15	Deletes compiled templates not used for 30 days
rangefront:refresh-route-matrix-cache	Daily at 03:00	Drops expired route legs and extends those next week's jobs still need
rangefront:monitoring-alerts:deliver	Every 15 minutes	Delivers alert notifications and escalations
rangefront:monitoring-alerts:expo-receipts	Every 15 minutes	Reconciles push receipts
rangefront:monitoring-alerts:resend-failures	Every 15 minutes	Resends alerts after transient failures
horizon:snapshot	Every 5 minutes	Horizon dashboard metrics
sanctum:prune-expired, queue:prune-failed	Daily	Removes expired API tokens and failed jobs older than 30 days

A heartbeat task also runs every minute and records the time. The health check reads it.

How they behave:

• Overlap. Each has withoutOverlapping with an expiry, so a crashed run frees its lock in 30 to 180 minutes, not a day. Each also has onOneServer, which needs a cache store shared by your servers (Redis).
• Timezones. Work that touches customers runs every hour, and each business is handled in its own timezone. Reminders, quote follow-ups and statements only go out between 8am and 8pm local (statements and invoice reminders from 9am). Statements go on the business's chosen day, or the last day of a shorter month. Quotes still expire overnight.
• Running twice. Every command is safe to run again. Messages carry a dedupe key, council notices are recorded in scheduled_notice_log, and a plan date or system due date only ever makes one job, even if its draft was cancelled.
• One business failing. Each business is handled on its own. If one throws, the error is logged and reported, the rest carry on, and the maintenance commands exit with an error so you can see it.
• Summary. Each maintenance command logs a one-line summary of what it did.

Each command takes --organization=<id or slug> where it applies to one business at a time, which is handy for testing. rangefront:generate-due-jobs also takes --force to run outside the 5am hour. Plugins can add their own scheduled work.

Draft jobs from plans appear in Drafts for dispatch to schedule. Someone can still choose Generate on a plan to make visits further ahead. See schedule.

Other commands worth knowing:

Command	Does
rangefront:install-defaults <org> [--enable-messages]	Installs the commissioning form and standard messages
rangefront:tenant:create, :list, :suspend, :resume	Create and switch off businesses. See above
rangefront:openapi [--output=path]	Writes the OpenAPI contract
rangefront:plugin:list, :make, :install, :enable, :disable, :uninstall	Plugin lifecycle. See extending
rangefront:hooks:docs [--check]	Regenerates or checks the hook catalogue in docs/extending.md

Health check

GET /up checks the install and returns JSON:

{"status":"ok","checks":{"database":{...},"redis":{...},"queue":{...},"scheduler":{...}}}

Check	Fails when
database	A query can't run
redis	Redis doesn't answer (only checked if the queue, cache or sessions use it)
queue	The queue can't be reached. More than HEALTH_QUEUE_MAX_BACKLOG jobs waiting (usually Horizon is down) only makes the status degraded, so a backlog never takes web servers out of rotation
scheduler	The heartbeat is older than 5 minutes, which means cron isn't running schedule:run

status is ok, degraded or down. down returns HTTP 503, so point your load balancer or uptime monitor at it. degraded returns 200 and means a check has nothing to measure yet, or the queue backlog is over its limit. The scheduler shows unknown until its first heartbeat, and stays that way if cron was never set up, so alert on degraded as well in production. The response never includes error text.

Storage

Photos, signatures, imports and generated PDFs go to object storage. The browser and the phone upload straight to the bucket with a short-lived signed address (RANGEFRONT_UPLOAD_URL_TTL), and the server then verifies the file's hash and scans it. The bucket therefore has to accept PUT requests from your console's address and from the field app, so set a CORS rule for it. Objects live under organizations/{organization_id}/.... Keep the bucket private, turn on versioning and server-side encryption, and see backup and restore.

If you use FILESYSTEM_DISK=local, the signed local fallback is meant for development.

Laravel also needs storage/ and bootstrap/cache/ writable by the PHP user.

Mail

Out of the box, mail goes through the MAIL_* settings, and MAIL_MAILER=log just writes it to the log, which is right for development and wrong for production. Either set a real SMTP server here, or have each business connect its own provider under Admin > Email and SMS. Business-level settings win. Texts work the same way with TWILIO_*.

Mail needs the queue worker. Customer messages are queued on notifications.

Checks after installing

1. GET /up returns "status":"ok". Give the scheduler a minute or two to write its first heartbeat.
2. Horizon shows workers running (php artisan horizon:status), with all three supervisors in its dashboard.
3. php artisan schedule:list shows the commands above.
4. Generate a PDF for a job and check it appears.
5. Send a test email and text from Admin > Email and SMS.
6. Open the console, sign in as the owner you invited, and check the live-update dot goes green.
7. Request a password reset from the sign-in page and check the email arrives (it uses the same mail settings).