Thirty Fake Outages Later, Only One Free Status Page Deserved Customer Trust
Photo: laptop showing uptime status dashboard in small office server closet, via pisces.bbystatic.com
A status page is a promise. When it says all systems operational, your customers stop checking. When it says nothing while their email is down, they start calling. At FreeSoft Labs, we wanted to know which free, open-source status page actually keeps that promise for a budget-conscious IT shop, so we rented a $6-a-month VPS and spent 30 days lying to three of them.
The box and the lineup
The host was nothing special: 1 vCPU, 1 GB of RAM, a 25 GB SSD, Ubuntu 24.04, and Docker. Total cash outlay for the month was six bucks. We deployed three free tools side by side:
- Cachet — the veteran incident-communication page. You post incidents, schedule maintenance, and publish updates. It does not watch anything for you.
- Uptime Kuma — a monitoring tool with a built-in public status page. It checks endpoints and shows uptime history.
- Gatus — a config-as-code health dashboard. You write YAML, it checks services, and it exposes a status view.
All three are open source. All three fit on the VPS. Only one of them made it to the end of the month without making us look foolish.
| Tool | What it actually does | Setup time | Native monitoring? |
|---|---|---|---|
| Uptime Kuma | Monitors endpoints and publishes a status page | 22 min | Yes |
| Gatus | YAML health checks with a status view | 38 min | Yes |
| Cachet | Manual incident posts and maintenance scheduling | 3 hr 40 min | No |
Cachet's setup time is not a typo. The community Docker image is aging, the PHP extensions fought us, and the queue worker needed a supervisor. If you enjoy Composer errors at 11 p.m., Cachet is a good time. If you run a four-person MSP in Ohio, it is a project.
The month of lies
We staged 30 events over 30 days: 12 service kills, 8 network blocks, 6 maintenance windows, and 4 slow degradations. We logged the real start and end times with a stopwatch and compared them to what each page published. This was not a load test. It was a honesty test.
Week 1: the ISP outage that was not
We blocked outbound traffic on the lab router for 11 minutes to simulate a dead business line. Uptime Kuma noticed in 48 seconds and sent an email and a webhook. Gatus noticed in 31 seconds. Cachet noticed nothing, because Cachet does not monitor anything. We logged into the admin panel and posted a manual incident nine minutes after the outage began. By then, two fake clients had already emailed the fake help desk.
That is the core problem with a manual status page: the page is only as fast as the person who remembers to update it. On a good day, that person is in a meeting.
Week 2: the disk that did not die
We killed a service container and left it dead for 17 minutes. Uptime Kuma's public page flipped to down, showed the failing check, and recorded the outage window within a minute of the real event. Gatus did the same, with a cleaner internal view. Cachet again stayed green until we typed the incident ourselves. We also forgot once. That is not a hypothetical. We literally forgot to post an outage on a page whose only job is posting outages.
Week 3: maintenance windows
Cachet wins this one, and it is not close. Scheduled maintenance is built into the product, with a start time, an end time, and a public message. Uptime Kuma has maintenance windows that silence alerts and mark the page accordingly, which is enough for most small shops. Gatus handled maintenance by flipping a config flag and redeploying. It worked, but there was no customer-facing calendar entry, so the page looked broken while we were deliberately breaking it.
Latency, accuracy, and the green lie
Here is the uncomfortable truth about free status pages: a false green is worse than a red. A red page tells customers you know. A green page during an outage tells them you do not.
| Tool | Detection speed | Recorded duration vs. actual | Customer-facing incident notes |
|---|---|---|---|
| Uptime Kuma | 48 sec avg | Within 1 min | Basic |
| Gatus | 31 sec avg | Within 30 sec | None |
| Cachet | Manual | Off by 6 to 11 min | Excellent |
Gatus was the fastest and the most accurate, but it is a health dashboard, not an incident-communication tool. It does not give you a place to write 'we are investigating a fiber cut in Dayton.' For internal use, that is fine. For customers, it is a green light with no context.
Email deliverability bit us on day one. Gmail throttled the first alerts until we fixed SPF and DKIM on a dedicated subdomain. That is not the tool's fault, but it is part of the real cost. SMS through Twilio worked fine and cost us 71 cents for 90 messages across three phone numbers. The setup took 40 minutes.
The invoice
Here is where the free tools stop being free. We tracked cash and labor separately.
| Line item | Cash cost | Notes |
|---|---|---|
| VPS | $6.00 | 1 vCPU, 1 GB, 30 days |
| Domain | $1.00 | $12 per year, amortized |
| TLS | $0.00 | Let's Encrypt |
| SMTP relay | $0.00 | Free tier, 100 messages per day |
| SMS via Twilio | $0.71 | 90 messages at $0.0079 each |
| Labor | $453.75 | 8.25 hours at $55 per hour |
| Total | $461.46 |
The labor number is not padding. Setup ate 4 hours and 40 minutes, mostly Cachet. Incident writing, DNS, email authentication, and the fake-outage testing ate another 3 hours and 35 minutes. If you do not count your own time, you are not doing a cost analysis. You are doing a hobby.
The cash-only cost was $7.71 for the month. That is a great story until you realize the manual tool moves the cost to on-call attention. Every outage you forget to post is a customer trust withdrawal.
What we kept
We kept Uptime Kuma. It is not perfect. The public page is a little plain, and the incident notes are basic. But it monitors, it publishes, it handles maintenance windows, and it does not require a PHP stack from 2019 to stay alive. For a small US IT shop that needs customers to trust the page when something really breaks, that is the right trade.
We would use Gatus again for an internal status board, especially if the team already lives in Git. We would use Cachet only if public incident prose is a hard requirement and someone has four hours to babysit the install. For everyone else, the $6 VPS plus Uptime Kuma is the honest answer.
The other two are gone. The page now says what it means.