Upptime alternative for checks more often than every five minutes
Upptime is capped by GitHub Actions cron, which will not schedule below five minutes and can run late when the runners are busy, so Logdash runs the checks on its own schedule instead - though Upptime stays free at any scale and Logdash does not.
Upptime is a good trick. A scheduled GitHub Action curls your URLs, commits the result, opens an issue when something is down and closes it when it recovers. No server, no bill, and the whole history sits in a repo you own. For a side project that is hard to beat, and you should not switch off it out of taste.
It runs out of road in two places. The first is the schedule. GitHub Actions cron does not accept anything finer than five minutes, and scheduled workflows are queued rather than guaranteed, so during busy periods they start late. Real detection time is five minutes plus whatever the queue is doing. A two minute outage can pass without a single failed check, and you find out from a customer.
The second is the repository. The workflows commit their results back and every incident is a GitHub issue, so monitoring noise lands in the same inbox as pull requests, and response time over the last month is spread across thousands of commits.
Logdash runs the checks itself, every minute on Builder and every 15 seconds on Pro, and keeps 90 days of hourly history without committing anything. Two things Upptime has that Logdash does not: it is free at any scale, and it is genuinely yours. Logdash has a free Hobby plan with paid tiers above it, and while it is AGPL-3.0 licensed, production self-hosting is not a one-command install yet.
A status page you still deploy yourself
npx create-next-app@latest status-page \
--example https://github.com/logdash-io/logdash.io \
--example-path templates/status-page-next
cd status-page
cp .env.example .env.local
# set LOGDASH_STATUS_PAGE_ID in .env.local, then
npm run devThat is the Logdash Next.js starter: one page, one component and a Tailwind theme you own, in your own repo. It renders on the server, refreshes at most once a minute, and its README has a one-click Vercel deploy. Logdash runs the checks and serves the data. The page, its styling and where it is hosted are yours, as they were with Upptime.
- Copy the sites list out of .upptimerc.yml Each url becomes one HTTP monitor, checked every minute on Builder or every 15 seconds on Pro instead of the five minute floor.
- Turn on the public status page Add the monitors to a page and point your own domain at it, or deploy the starter above. Either replaces the GitHub Pages site.
- Attach Telegram Connect a Telegram channel to each monitor. The next time an endpoint returns a 5xx, the alert is in your chat instead of being an issue notification you will read tomorrow.
Logdash vs Upptime
| Feature | Logdash | Upptime |
|---|---|---|
| Minimum check interval | 1 minute on Builder, 15 seconds on Pro | Five minutes, and Actions cron can start late |
| Cron and background jobs | Push heartbeats on Pro, one ping per check window | Not covered, Upptime pulls URLs |
| Logs and metrics | Same dashboard, eight SDKs | Uptime only |
| Cost | Free Hobby plan, paid tiers above it | Free, on your own Actions minutes |
| Self-hosting | Local development only today | Runs entirely in a repo you own |
| Where incidents live | Failed checks in the uptime history, no incident posts | GitHub issues, commentable and searchable |
| Status page | Hosted page on your own domain, or your own build via API | GitHub Pages site you style and deploy |
When Upptime is the better pick
- The bill is zero and it stays zero. If cost is the constraint, nothing here competes.
- Everything lives in a repo you control. No account to lose, no vendor to outlive.
- Five minutes is fine for most side projects. If nobody is woken up at 3am, the floor does not matter.
- You want no outside data source at all. Logdash lets you style and deploy the page yourself, but the data still comes from Logdash. Upptime keeps the checks, the history and the page in your own GitHub organisation.
What moving actually looks like
Keep the repo. Recreate the sites list as monitors and run both for a week, which is usually enough to show you the gaps where the five minute grid missed something. When you are convinced, disable the workflows and archive the repo rather than deleting it. Nothing imports your old issues, so that archive is your incident history.