Skip to content

How it fits together ​

A Replora installation is one Laravel application with a React dashboard, a PostgreSQL database, a folder or bucket for recordings, a background worker and a cron line. This page shows how they work together, which is useful when you plan a server, a backup or a move.

5 minutesSite owners, DevOps, developersSingle user · SaaS

The big picture ​

 Visitors' browsers ─► your website ─► <script async src="https://analytics.example.com/t.js" data-site="KEY">
                                          ├─ GET  /c/KEY            the site's rules (masked and left-out elements)
                                          ├─ POST /i/KEY            compressed event batches, every few seconds
                                          ├─ GET  /s/KEY · POST /s/KEY/ID   surveys and their answers
                                          └─ GET  /x/KEY            A/B test variants
                                                   │
                                                   ▼
                              queue (RECORDINGS_INGEST_QUEUE) ─► queue:work ─► recording storage + database

 Your team ─────────► dashboard (/sites, /system …) ─► /api/v1/…   (session sign-in)
 Scripts, apps ─────► /api/v1/…                                    (API tokens)
 Background ────────► queue worker (recordings, alerts, announcements, HubSpot) · scheduler (cron)

A recording's journey ​

  1. The snippet on your website loads t.js from your Replora address (APP_URL). The data-site attribute is the website's public key.
  2. The rules: the script asks /c/<key> which elements to mask or leave out (the website's privacy settings), so they never leave the visitor's browser.
  3. Batches: while the visitor browses, the script sends gzipped batches of events to /i/<key>. The endpoint only checks them (known and recording website, an allowed domain, the website's recording rules, size limits of 5 MB compressed and 50 MB unpacked) and puts them on the queue. It answers at once, so ingest stays fast under load.
  4. Processing: the background worker (php artisan queue:work) unpacks each batch, writes the events to recording storage, and updates the session, its pages, heatmaps, form, error, funnel and insight data in the database. Alerts that match are sent from here too.
  5. Replay: the dashboard loads the session's events from recording storage and rebuilds the page in the player. Copies of the stylesheets the page used are kept (/css/<sha>.css), so replays keep their look after the website changes.

No worker, no recordings

If no worker runs, batches wait in the queue and new sessions don't appear. The Background jobs check on the System page fails when jobs wait longer than five minutes. See Background worker & email.

The public endpoints (/t.js, /i/…, /c/…, /s/…, /x/…, /css/…) use no session, no cookies and no CSRF token; they send their own CORS headers, so any website can call them. They keep working while the dashboard is in maintenance mode.

Where things are stored ​

DataWhere
Accounts, users, team roles, invitations, API tokensDatabase: organizations, users, user_sites, invitations, personal_access_tokens
Websites and their settingsDatabase: sites
Sessions, page views, tags, shares, saved filtersDatabase: recording_sessions, recording_pageviews, recording_chunks (the replay index), recording_tags, recording_shares, recording_saved_searches
Recorded events (what the player replays)Recording storage, one folder per session (<site id>/<session id>): storage/app/recordings or your S3 / R2 / MinIO bucket. See Recording storage
Heatmaps, forms, funnels, errors, events, insights, networkDatabase: heatmap_*, form_*, funnel_*, error_*, custom_events, insight_*, network_*
Surveys, A/B tests, alerts, AI summariesDatabase: surveys, survey_responses, experiments, experiment_*, alerts, alert_events, session_summaries, site_digests
SaaS: plans, subscriptions, pages, affiliates, ticketsDatabase: plans, plan_*, stripe_* or paddle_*, pages, affiliate_*, ticket_*
Settings changed in the dashboardDatabase: replora_state (secrets encrypted with APP_KEY)
Uploaded logo, favicon, attachmentsstorage/app/private
Installation settings.env
The verified purchase codestorage/app/replora-license.json
"Installed" marker (closes the installer)storage/app/replora-installed.json
Update backups (previous version's files)storage/app/updates
Logsstorage/logs/laravel.log

Keep APP_KEY safe

Saved secrets (SMTP password, API keys, payment keys) are encrypted with APP_KEY from .env. Back it up together with the database: without the same key they can't be read. See Backups.

Modules ​

Every feature area is a module: a folder in modules/ (Recordings, Heatmaps, Insights, Funnels, Forms, Errors, Events, Alerts, Surveys, Experiments, Summaries, HubSpot, Announcements, Help and Installer). The SaaS edition adds saas/modules/ (Plans, Signup, BillingStripe, BillingPaddle, Landing, Pages, Affiliates, Tickets).

The release zip contains the application and the Installer module. The other modules/ folders are the core pack: the installer downloads them from the license server after your purchase code is accepted, checks their signature and puts them in place. Updates refresh them the same way. See Your purchase code and Architecture & modules.

Background work ​

WhatRunsDoes
Queue workerphp artisan queue:work, kept running by Supervisor or systemdProcesses recording batches, keeps stylesheet copies, sends alerts, announcement emails and HubSpot updates. See Background worker & email
Schedulerphp artisan schedule:run from cron, every minuteDeletes expired recordings (03:00), checks for updates (twice a day) and services (every five minutes), sends alert digests, funnel regressions, AI site digests and (SaaS) trial reminders, ticket reminders and affiliate payouts. See Scheduler
Live updatesPolling by default; optionally Laravel Reverb (php artisan reverb:start)Keeps live visitors and lists fresh in the dashboard. See Scaling

Replora: see what your visitors see.