This relay is working normally.
Everything below was measured when you loaded this page. There is no switch an operator can leave set to green: if a check fails, this page says so the next time somebody looks, whether or not anyone has noticed yet.
What it is carrying
The figure that matters for capacity is how many phones are holding a connection open at the same moment, not requests per second, because a request that returns instantly and one that waits half a minute are one request each and cost very different amounts.
| Connections held right now | 0 |
| Most ever held at once | 2, reached 2026-10-20T00:15:14Z |
| Connections held since this process started | 104 |
| This process has been running for | 5 days |
No ceiling has been set on this relay, so this page can tell you the load but not whether it is a lot. We would rather say that than print a plausible-looking limit: a held connection does not tie up a database connection, so the obvious candidates are not the answer, and a guess carrying a status page's credibility is worse than an admission. The peak resets whenever the process restarts, a deploy makes it look like traffic vanished, which is why the uptime is printed next to it.
Who it is carrying
The honest size of this relay. Counts only, taken from the database when the page loads: an account or a device is active if the relay heard from it in the last thirty days.
| Accounts | 12 |
| Accounts active in the last thirty days | 10 |
| Devices active in the last thirty days | 14 |
Small numbers are printed as small numbers. A relay that rounds eight people up to "hundreds" has told you what its other figures are worth.
What it depends on
| Component | State |
|---|---|
| Database | Answering |
| Attachment storage | Answering |
A row that says it was not checked means exactly that, and is not a quiet pass. Object storage sits behind another company's API, and probing it every time a stranger loads a public page would run up the operator's bill and hand anyone a way to generate traffic against their bucket, so it is not probed, and this page tells you rather than showing you a green tick for something nobody looked at.
Reading these numbers yourself
The same reading is available as JSON at
/status.json, and
/health answers in one word for a monitor
that only wants up or down.
/status.json answers 200 even when something is
broken, with the trouble in the body, a page that returns an error during an
outage shows you a blank screen at the one moment it had something worth
saying. /health is the endpoint that exists to be a status code,
and it reports a deploy in progress as draining rather than as a
failure, so handing over cleanly does not look like going down.
Counts only, never names. Nothing here records which account or which device is connected, and no version of this page will: the number of people connected is safe to publish and the list of them is not, so the list is never assembled in the first place.