Minecraft hosting uptime, measured every minute
Everything we measure is answering right now. One part had downtime in the last day: Your Minecraft server. Below is every part of the service on its own line, with what it scored over the last 23 days and how often we check it. The method is written out underneath, so you can judge the figures rather than take a single number on trust.
Every part of the service
One row per part, because they also break separately: signing in can be down while every server keeps running, and then it should say so here. The dot is the state at the last check, the percentage is the share of our checks that answered over the last thirty days.
| Part of the service | State | Answered | We check |
|---|---|---|---|
| Your Minecraft serverWhether the servers we host were actually running and answering, the thing your players notice first. | Up, recent downtime | 98% | every five minutes, per server |
| Play addressesThe layer your players connect through, so that the address you handed out leads somewhere. | Operational | 100% | every minute |
| Server platformThe layer that starts, stops and runs the container your server lives in, and that the panel talks to. | Operational | 100% | every minute |
| Control panelA real request that has to reach the database, so this is green only if the panel can truly show your server. | Operational | 100% | every minute |
| Account areaYour own pages: servers, invoices and settings, checked from the outside like any visitor. | Operational | 100% | every minute |
| Sign inThe part that has to answer before anyone can get through the login screen. We sign nobody in to test it. | Operational | 100% | every minute |
| OrderingThe page where a plan is put together and paid for, checked on a real plan and not on a test page. | Operational | 100% | every minute |
| WebsiteThe pages you are reading now, requested from a machine that is not the one serving them. | Operational | 100% | every minute |
| Nightly backupsEvery backup we take for you, counted as taken or failed, with no second attempt hidden from the figure. | Operational | 100% | every night, per server |
Incidents we reported
Everything we post on the status page appears here too, in the same words, so you do not have to watch two places. The status page is hosted away from us on purpose, so it keeps working when we do not.
| Date | What we reported | Length | State |
|---|---|---|---|
| 10 September 2026Between 00:00 and 01:00 UTC on 10 September, users experienced errors when signing in to the account area and the panel. Existing sessions and all other services remained operational. The issue has been resolved. | Temporary issues affecting login | stated in the report | Resolved |
| 22 August 2026We are seeing elevated error rates on some game servers. Our team is looking into it now and we expect this to be resolved within the hour. Your worlds and data are safe. | Elevated errors on some game servers | 3 hours 2 minutes | Resolved |
Per month
All our checks of that month together, including the bad days. A month that has only just begun shows the days it has.
| Month | Answered |
|---|---|
| September 2026 | 98% |
| August 2026 | 99.9% |
How we measure it
Every server we host is checked around the clock, and one measurement per server is kept every five minutes. The website, the panel with your login and the platform that runs the containers are requested once a minute from a machine that is not the one being checked. The panel check is a real request that has to reach the database, so it is only counted as answered if you could actually sign in and see your server.
A server that you stopped yourself is not downtime and is left out. A server that was off for a whole day does not count that day at all. What remains is the time your server should have been running. Raw measurements are kept for forty days, the daily totals stay, and nothing on this page is rounded up: a percentage is cut off at one decimal rather than rounded in our favour.
A server that went down on its own counts as well, including a crash caused by a plugin or a modpack the customer installed. We could leave those out and the figure would look better, but the minutes are real for whoever was playing. Our watch looks every fifteen seconds and starts a server that stopped on its own, so most of those minutes are a handful at a time rather than an evening.
You will not find the number of checks behind a percentage here either, and that is deliberate: a server is measured every five minutes, so the count would say exactly how many servers we run. The share that answered says what you need to know, and the rate we check at is written next to it.
What you will not find here is the number of servers or machines we run, or where any customer sits. Those are not ours to publish. What we can show is every check we made and what came back.
Anomalies
Nothing to report for this period. When a check fails twice in a row we write it down, and once we know what it was it appears here in plain language, with the date and how long it lasted. Live incidents are posted on the status page while they are happening.
Questions about these figures
What exactly do you measure?
Nine things, all of them ours, each on its own line above. Every server we host is checked around the clock and a measurement is kept every five minutes. The website, the control panel, your account area, signing in, ordering, the platform that runs the containers and the layer your players connect through are each checked once a minute from a machine that is not the one being checked. Every backup we take for you is counted as taken or failed.
Does a server that I switched off myself count against you?
No, and that is the whole reason these numbers mean anything. A server that you stopped yourself is left out, and a server that was off for a full day does not count that day at all. What is left is the time a server should have been running and was not.
How is this different from the status page?
The status page is where we post while something is going on, and it is hosted away from us so it keeps working when we do not. This page is the history: what our own checks recorded over the past days and months, with the method next to it. Anything we post on the status page shows up here as well, in the same words, so you only have to watch one of the two.
Why do you not simply promise one hundred percent?
Because nobody can keep that promise, and a number you cannot check is worth nothing. We publish what we measured, including the bad days, and the service credits in the agreement are what you can hold us to.
How often is this page updated?
The previous day is closed off every night and the page follows within minutes. Raw measurements are kept for forty days, the daily figures stay, so the history on this page only gets longer.
Where to go next
What we owe you when something does go wrong, including service credits, is written out in the service level agreement. What is happening right now is on the status page. The machines behind these numbers, and the reason a server with fewer players can still feel slow, are covered on high performance hosting and in the guide on reducing lag. If your own server is not behaving, support is the fastest way to reach us.
Delivery: most servers are handed over instantly and run within seconds of payment, the rest follow within 60 minutes during the day and within 12 hours at night.