The check passed
The component answered as expected at the time shown. It says nothing about the moment after that.
Service status
Is this instance answering?
A live snapshot of the instance that served this page. Each part is reported separately: that the web server answered, that the application process is up, whether the database answers a query, and whether the web app itself was found. This is one instance at one moment. It is not a project-wide status board, and it cannot see services the server does not expose.
Checking this instance...
Running the checks now.
| Component | What it is | State |
|---|---|---|
| This website | The public page you are reading, served by the instance over HTTP. | Checking |
| Application API | The process that answers requests from the app and from this site. | Checking |
| Database | Where accounts, communities, messages and permissions are stored. | Checking |
| Web app | The client the server serves when you open the app. | Checking |
| Voice relay | The STUN and TURN service this instance runs, at the address its clients are given. | Checking |
Times are shown in your browser's local time.
What has happened here
Nothing has been reported.
What the checks show
What a green check does and does not mean
A passing check is deliberately narrow. It confirms one thing answered, not that everything works.
The API check proves a process answered
It calls the instance's own health endpoint. A reply means the API process is running and can answer a request. It does not prove that every dependency, background job or feature is healthy.
The database check proves one query returned
It runs a single read against the database. That catches an unreachable database, which is the failure that stops everything. It is not a check of every table or every write path.
It cannot see what the server does not expose
This instance publishes health for the API and the database only. The realtime gateway, mail delivery, uploads and voice have no individual check here, so this page stays silent about them rather than guessing.
A failure can also be the connection to the source
A check that fails means the checker could not complete it, which can be the instance or the network between the two. When your browser ran the check, reload before concluding the instance is down.
The states
What each state means
It answered, but a part is missing
The component responded and something it depends on is not in place. The clearest case here is a server that runs the API but was never given the web app to serve.
The check did not pass
The component was reached but did not answer as expected, or refused. This is a real failure, not a warning.
The check could not run
The request never completed. That can be a network problem, a blocked request, or an instance that is not there. No conclusion is safe, so none is drawn.
When something is wrong
What to do next
- Reload this page and run the checks again. A single failure can be a dropped connection.
- If the website is operational but the API or database is not, the problem is on the server, and nothing you change locally will help. If it is your instance, see the documentation; if it is not, its operator is the person who can fix it.
- If every check is unknown, suspect the connection between you and the instance first. Try another network or a different device.
- Still down, or you have a detail this page cannot show? The support page explains who to tell, and the project's Discord is where the community talks.
How far this record goes back
When the instance runs the bundled status monitor, the results above are that monitor's most recent checks, held in memory for a short window and reported with the time each one was taken. Restart the monitor and the window begins again, so it is a recent snapshot, not a permanent history. Without the monitor, your browser runs each check once and nothing is stored.
The timeline below is different: it is what an administrator of the instance wrote down, and it stays until the instance's operator removes it. Where there is nothing there, the instance has nothing recorded — not that everything has been working, because nobody wrote that down. There is no uptime percentage, because nothing here computes one and a number that was not measured is not worth showing.