reachability
A real call, every minute
A real RPC request against each chain, with an API key, over the public endpoint — exactly like a customer call.
Status
Measured, not asserted.
Every number on this page comes from a real call made a minute ago, not from a health endpoint reporting on itself. The gateway is in private beta and there is no SLA yet — but the measurement is real, and so is the incident history.
| Service | State | 24 h | 7 d | 30 d | p50 | p99 | Lag |
|---|---|---|---|---|---|---|---|
| Fetching /api/status.json… | |||||||
Networks and backends
Per chain: how many backends are answering, whether a Phantrium-operated node is among them, and the block height we serve from. A backend that answers fast with stale data is broken, so height is checked, not just reachability.
| Network | Chain ID | Backends | Served from | Block height | Checked |
|---|---|---|---|---|---|
| Loading… | |||||
Latency per method, from real traffic
Averages hide the calls that hurt. These are percentiles over a sample of actual customer calls in the last 24 hours — cache hits and our own monitoring traffic excluded, so the numbers are the honest ones rather than the flattering ones.
| Method | Network | Calls | p50 | p95 | p99 |
|---|
No customer traffic in the last 24 hours
The gateway is in private beta, so this table is empty until someone other than us calls it. It fills itself; there is nothing to switch on.
How these numbers are made
So you can judge them, and argue with them.
freshness
Block-height lag
How far behind the chain head our serving path is, in blocks. A node that answers instantly with stale data is broken, and this is the number that catches it.
honesty
Two vantage points
Every round runs twice: once over the public hostname through Cloudflare, once straight at the service. When only the public path fails, we say so instead of guessing.
A window only gets a percentage once half of its measurements exist.
Uptime is measured on the public endpoint.
Both vantage points sit on the same machine as the service.
Incidents
Every incident that affects a customer is published here with cause, impact and the fix — including the ones caused by us. An entry appears the moment the measurement confirms it, before we have an explanation; the explanation is added afterwards and the entry stays.
Loading incident history…
Want to be emailed when a service you use has an incident? Turn it on in the dashboard, or say so in the contact form — one message when it starts, one when it is over.