Errors Under Control We're Building the System That Watches Your Site

Hotel WD is building an error-monitoring layer (Sentry) into its core. The goal: if your site stumbles, we see it before you hear about it from a guest — and fix it without you having to open a ticket.

Errors Under Control: Hotel WD Is Building Its Real-Time Monitoring System

A traveler tapped "Reserve". The page returned something they didn't expect. They saw a strange dialog, closed the tab, and started searching for another property. Did you ever find out it happened?

For most hotels the honest answer is no. Visitors rarely fill out a feedback form, rarely call the front desk, rarely send an email about something that "almost worked". They close the tab. The owner sees the symptom weeks later — a soft drop in monthly bookings, no clear cause — and the revenue is already gone.

This pattern has a name in software engineering: silent failure. It is the quiet enemy of every website that runs at scale.

This week we started wiring Sentry into the core of Hotel WD. The scaffolding is in place and the rollout is underway. The goal is simple: whenever a guest experiences an error anywhere on your site, we want to see it before you — or the guest — has to report it.

Sentry real-time error pipeline: Hotel WD app captures the exception, Sentry enriches it with stack trace and breadcrumbs, the engineer is alerted on Slack.

What Sentry Is, in Plain Terms

Sentry is the production observability platform used by GitHub, Microsoft, Disney, Spotify and over 100,000 other engineering teams. Founded in San Francisco in 2012, it operates a European data region in Germany — which matters for GDPR-aware operators.

Its single job is to capture, in real time, every error and performance degradation that happens inside live software, and route the right context to the engineers who can fix it.

Every hotel on Hotel WD will be part of that pipeline as the rollout completes. There is nothing for you to install, configure, or pay extra for — it all happens behind the scenes.

The Old World vs. The New One

Old world. Something on your site broke. A guest closed the tab. The next reservation was lost. Your monthly report showed "softness" in bookings; the cause stayed invisible. The damage was silent and cumulative.

New world. Something on your site breaks. The monitoring pipeline records it — the exact error, the exact line of code, the guest's path leading up to it, and the device they were using — and surfaces it to us. That context is what lets a fix ship quickly instead of after weeks of guesswork. The aim: catch the dip before it ever shows up in your monthly report.

This is how Amazon, Booking and Airbnb have run for years. We've taken the same operational discipline and brought it down to the scale of one hotel.

What Sentry Captures When Something Breaks

The moment an exception is thrown on your site, Sentry collects:

  • The full stack trace — which file, which line, which call chain triggered the failure.
  • Guest context — browser, device, country, the page they were on, the room and date range they were looking at.
  • Breadcrumbs — the exact sequence of pages and clicks that led to the error, like the pre-flight conversation in an aviation black box.
  • Release information — which Hotel WD deploy this issue first appeared in. This is decisive: it tells us in seconds whether something is a regression or a brand-new bug.
  • Frequency pattern — is this happening to fifty visitors or just one? On every browser or only Safari on iOS 17?
  • You don't have to read any of this. Our engineers do. You experience the result: your site keeps working.

    What Else Landed in This Sprint

    Sentry alone isn't enough; the surrounding infrastructure has to be sound. The other changes we shipped this week:

    1. Multi-Runtime Error Capture

    The @sentry/nextjs SDK is now embedded in the Hotel WD core, with separate channels for the Node.js server runtime, the edge runtime, and the browser. Personally identifiable information (PII) is off by default — only anonymous error traces and technical context are sent. GDPR-aligned, ready for hospitality compliance reviews.

    2. Health Check Endpoint (/api/health)

    A new endpoint pings the database every second to confirm the application is alive. PostgreSQL responds, the endpoint returns 200 OK; it stalls, and we get 503 Degraded. This lets us:

  • Restart the container automatically when it dies.
  • Slot in a load balancer later that won't route guests to a failing server.
  • Operate blue-green deploys with proper readiness signals.
  • 3. Cache Coherence (Cloudflare TTL Sync)

    The TTL gap between Cloudflare's edge cache and Next.js's internal cache has been closed. The theoretical worst case for a stale price was seven days; we have shrunk that window to 24 hours. Room availability and rate updates now propagate within one day, every time.

    4. Cumulative Layout Shift Fix

    Our cookie consent banner used to "pop in" during page load — a subtle but real source of layout shift that Google PageSpeed flags and search ranking algorithms penalise. We refactored the banner to render in place from the first paint, becoming visible only via opacity. Layout shift goes from measurable to zero.

    5. Build Pipeline Hygiene (Docker)

    Our production Docker build context dropped from roughly 600MB to 37MB. Deploys are 1–2 minutes faster, server bandwidth is lower. The container also now runs its own HEALTHCHECK directive every 30 seconds.

    6. Concurrent Deploy Protection

    A race condition existed where two GitHub Actions deploys could begin in parallel. Now closed with a concurrency group.

    What This Means for the Hotelier

    You don't need to know the technical depth above. But if you operate on Hotel WD, you should know it is being applied on your behalf.

    Hotel WD's commercial promise reduces to a single sentence:

    "Don't deal with the technology side of hospitality — we will."

    That promise can't be kept by good page design alone. It has to hold when the page goes live, when an error fires, when latency spikes, when prices update, when a payment processes. Engineering discipline at every layer is the only way it works.

    The infrastructure layer we shipped this week is the concrete step from "one hotelier's personal website" to enterprise-grade SaaS.

    At a Glance

    Area
    Time to detect a failure
    Before
    Hours or days
    New Hotel WD Standard
    Automated capture as it happens
    What It Means For You
    The goal: we see it before you have to report it
    Area
    Failure context
    Before
    "Something broke"
    New Hotel WD Standard
    Stack trace + browser + click trail
    What It Means For You
    Fixes target the real cause, not guesswork
    Area
    Silent failures
    Before
    Untrackable losses
    New Hotel WD Standard
    Tracked and visible to us
    What It Means For You
    Monthly reports get honest
    Area
    Health check
    Before
    None
    New Hotel WD Standard
    /api/health polled every second
    What It Means For You
    Automatic recovery
    Area
    Cache coherence
    Before
    Up to 7-day staleness
    New Hotel WD Standard
    24 hours maximum
    What It Means For You
    Prices and inventory stay fresh
    Area
    Page CLS (PageSpeed)
    Before
    Layout shift penalty
    New Hotel WD Standard
    Zero shift
    What It Means For You
    SEO ranking advantage
    Area
    Build size
    Before
    ~600MB
    New Hotel WD Standard
    37MB
    What It Means For You
    Faster deploys, smaller error window

    What's Next

    The work shipped this week is a starting line, not a finish. Layers on the runway:

  • Real-time Slack and email alerts wired to Sentry — instant team notification.
  • Automated PostgreSQL backup with restore drill — zero data loss in disaster scenarios.
  • Two-factor authentication (2FA) — a critical security layer for hotel owner accounts.
  • GDPR data export endpoint — for guests requesting a copy of their personal data.
  • k6 load testing at 100 concurrent users — to measure system behaviour under traffic surges.
  • Public status page — operational health visible to customers in real time.
  • The Hotel WD roadmap is being executed with a level of transparency that is uncommon in the hospitality software market.

    Closing

    When a guest visits your site, we don't want them wondering whether things are working. We want them booking without the question entering their mind.

    Getting there means being certain everything is in fact working. And being certain means measuring it, every second, in production.

    Sentry is one of the foundation layers that earns that certainty. As of this week, it is part of the Hotel WD core. More layers are on the way.

    Talk to Hotel WD and bring your hotel onto an engineering-grade platform →

    Tahir Dinç
    AUTHOR

    Tahir Dinç

    Turkey27+ YEARS EXP

    I have been working in the digital field since 1999. I still hold the position of Digital Marketing Manager at Türk SEM. I have also been involved in tourism-related activities since 2005.

    Hotel WD·Strategy Team
    Errors Under Control: We're Building the System That Watches Your Site | Hotel WD - Hotel WD