# Where to host Uptime Kuma in 2026: five options, priced > Where to run Uptime Kuma: Ownkube, Fly.io, Railway, Render or your own VPS, what each costs, why Heroku is out, and where the monitor should live. - **Published:** 2026-11-03 - **Author:** Abhishek Jaiswal - **Section:** Compare (https://ownkube.io/compare) - **Tags:** uptime-kuma, self-hosted, monitoring, paas-comparison, ownkube-compute - **Canonical URL:** https://ownkube.io/compare/where-to-host-uptime-kuma - **Cover:** https://ownkube.io/blog/where-to-host-uptime-kuma.png --- Uptime Kuma runs well anywhere that gives it an always-on container and a disk that survives a restart. In 2026 the practical choices are Ownkube (about $5 a month, one click), Fly.io (from about $2.34), Railway (from the $5 Hobby fee), Render (from $7.25) and a VPS you run yourself. Heroku and Replit are out: Uptime Kuma's own README lists them as unsupported, because their filesystems don't keep the data file. Wherever you run it, keep it off the platform your apps run on, since a monitor that goes down with your apps can't tell you they are down. ## Our picks | If you want | Pick | |---|---| | One click, HTTPS and a persistent volume set up for you, on a prepaid bill | Ownkube | | The lowest line item and you are fine writing a `fly.toml` | Fly.io | | A template deploy on a platform you already use | Railway | | A flat price and you already run services on Render | Render | | Full control, and you already patch a server | Your own VPS | | No server at all | A hosted monitor such as UptimeRobot | ## What Uptime Kuma needs from a host Uptime Kuma is a single Node.js process. It stores monitors, check history and settings in a SQLite file, so the host needs four things: - A disk that survives restarts and redeploys. Uptime Kuma's README warns that network file systems such as NFS are not supported, so use a local or block volume. - A process that never sleeps. Free tiers that spin down on idle stop your monitors without telling you. - One instance. SQLite allows one writer, so you run one copy, not a scaled group. - A public hostname with HTTPS, for the dashboard and for status pages your users will open. ## The options at a glance Prices are approximate list prices from each provider's pricing page, as of November 2026, for one small instance with a 1 to 2 GB volume. | Host | Monthly cost | Disk | Setup | Who patches the server | |---|---|---|---|---| | [Ownkube](/templates/uptime-kuma) | About $5, metered per minute | 2 GiB volume, no storage charge | One click from the Marketplace | Ownkube | | [Fly.io](https://fly.io/pricing/) | About $2.34 (256 MB Machine plus 1 GB volume) | Fly Volume, $0.15 per GB | `fly launch` and a `fly.toml` | Fly.io | | [Railway](https://railway.com/pricing) | $5 Hobby fee, which includes $5 of usage | Volume, about $0.15 per GB | Template | Railway | | [Render](https://render.com/pricing) | $7.25 (Starter instance plus 1 GB disk) | Persistent disk, $0.25 per GB | Docker image service | Render | | Your own VPS | The server, plus your time | Local disk | Docker Compose | You | ## The options, one by one ### Ownkube Uptime Kuma is in the Ownkube Marketplace. One deploy creates the app from the official image, mounts a 2 GiB persistent volume for the SQLite file, and gives it an `*.ownkube.app` hostname with automatic HTTPS. There is no Postgres to pay for, since Uptime Kuma doesn't use one. It costs about $5 a month: CPU and memory billed per minute, and the volume carries no storage charge. It runs in Mumbai or Germany. You can add your own domain for the status page. Finish setup right after deploy, since the first person to open the dashboard creates the admin account. And leave sleep off: a sleeping app runs no checks, so your monitors would stop until someone opened the dashboard. Billing is a prepaid wallet. Pay as you go starts with a $5 top-up, and unused credit never expires. See the [Uptime Kuma template](/templates/uptime-kuma), or [deploy it now](https://app.ownkube.io/dashboard?deploy=uptime-kuma). ### Fly.io Fly.io has the cheapest line here. A `shared-cpu-1x` Machine with 256 MB is $2.19 a month running all month, and a 1 GB volume adds $0.15. You write a `fly.toml` that mounts the volume at `/app/data`, then run `fly deploy`. A large set of monitors needs more memory than 256 MB, which moves the line up. Volumes bill on provisioned size even while the Machine is stopped. Fly.io also runs in many regions, which helps with the placement question below. See [Fly.io pricing explained](/compare/fly-io-pricing-explained). ### Railway Railway has Uptime Kuma templates that attach a volume for you. The Hobby plan is $5 a month and includes $5 of usage. Memory bills at about $10 per GB a month, CPU at about $20 per vCPU a month, and volumes at about $0.15 per GB. A small instance can fit inside the included usage. A heavier one pushes past it, so check the usage page after the first week. See [Railway pricing explained](/compare/railway-pricing-explained). ### Render Render can run the official Docker image with a persistent disk. Disks need a paid instance, so the floor is a $7 Starter instance plus $0.25 per GB of disk. Render's docs note two catches. A service with a disk loses zero-downtime deploys, so each deploy briefly takes the monitor offline. It also can't scale past one instance, which doesn't matter for SQLite. See [Render pricing explained](/compare/render-pricing-explained). ### Your own VPS Uptime Kuma's own install guide is Docker Compose on a server, and it is the most common setup. You get full control and the lowest unit cost if you already run a box. You also own everything else: OS updates, the reverse proxy and certificates, firewall rules, disk backups, and noticing when the server itself has died, which a monitor running on that server can't do. ## Put the monitor somewhere else If Uptime Kuma runs on the same platform, region or server as the apps it checks, an outage there takes the monitor down too. Your status page goes dark while users are looking for it, and no alert goes out. That holds for Ownkube as much as anyone. If your apps run on Ownkube in Germany, put Uptime Kuma on a different provider, or at least in Mumbai. If your apps run on Render or Fly.io, Ownkube is a reasonable separate home for the monitor. The same logic applies to alerts. Send them somewhere that doesn't depend on the monitored apps, such as email, Slack, Discord or Telegram, all of which Uptime Kuma supports. ## When not to self-host Uptime Kuma If you only need a handful of HTTP checks and don't care about owning the data, a hosted monitor is less work. UptimeRobot's free plan covers 50 monitors at a 5-minute interval, with commercial use allowed. Its cheapest paid plan, Solo, checks every 60 seconds for $9 a month billed yearly. Self-hosting wins when you want 20-second checks without a per-monitor price, private TCP or keyword checks, branded status pages, or simply to keep the history yourself. ## A quick checklist before you choose - Confirm the data directory sits on a volume that survives a redeploy. - Turn off any idle sleep or spin-down. - Run one instance only. - Put the monitor on a different provider or region from your apps. - Decide how you will back up the data file. - Send a test alert to every channel before you rely on it. ## FAQ ### Can I run Uptime Kuma on Heroku? No. Uptime Kuma's README lists Heroku as unsupported. Heroku's filesystem is reset on every restart, so the SQLite file and all your monitors would be lost. ### Can Uptime Kuma run on a free tier? Not reliably. Free tiers that sleep stop the checks, and the ones that don't sleep usually have no persistent disk. Plan on a small paid instance with a volume. ### How much memory does Uptime Kuma need? Memory grows with the number of monitors and how often they check. Start small, watch the memory chart for the first week, and size up if it climbs. ### What does Uptime Kuma cost on Ownkube? About $5 a month: its CPU and memory billed per minute, with the 2 GiB volume included at no storage charge. Pay as you go starts with a $5 top-up. ## Where Ownkube fits Ownkube is a good home for Uptime Kuma when you want it running in one click with HTTPS and a persistent volume, and you are not watching apps on Ownkube in the same region. It runs alongside the rest of the Marketplace, and the same wallet pays for your apps and databases too. [Deploy my first app](https://app.ownkube.io/login). ## Search This site has a search endpoint: https://context7.com/api/v2/ask?siteKey=ask_c64dfe8ffaa8d3ad19576838&query=. It returns documentation that answers the question, with a source link for each part. No API key is needed. If nothing matches, it says so.