Endpoints
Find an endpoint, check its latest probe and copy the exact URL.
Directory. Every endpoint the registry knows about.
Endpoints
Searching the endpoint catalog · probe-derived
Fleet observations. The recorded probe coverage and routing context.
Loading fleet observations
Pools. The managed RPC pools, and how much of each can be routed to.
Point a client at the managed proxy and it routes to the best eligible member:
observed eligible ÷ recorded members · pool p50 is the median of reported member latencies · probe-derived
Latency. How long the last probe took, at the ends of the distribution.
last recorded latency per endpoint · measured endpoints in loaded search results · probe-derived
Incidents. What is failing, and what was.
Endpoint incidents
Loading recorded incidents · probe-derived
How pool eligibility is decided
An endpoint enters the pool on sustained health, not on a single successful check, and leaves it on sustained failure. That hysteresis is deliberate: routing traffic to an endpoint that answered once is how a flapping host becomes everyone's outage.
Latency is measured, not advertised
Times shown are round trips from our probes, so they describe the endpoint under real conditions rather than an operator's benchmark. Compare them as relative signals; absolute values depend on where the probe runs.
Raw identifiers & sources
- managed RPC proxy
https://api.metagraph.sh/api/v1/rpc/proxy- endpoint-incidents
https://api.metagraph.sh/api/v1/endpoint-incidents