API reference
Health and metrics
Unauthenticated probes for your orchestrator and monitoring. Put them behind your network, not the internet.
Liveness
GET/live
Auth: None.
Never touches a dependency, so a database blip doesn't restart every node.
Responses
{
"status": "ok"
}
{
"status": "shutting_down"
}
Example
curl http://127.0.0.1:3000/live
Readiness
GET/ready
Auth: None.
Fails on Postgres (unreachable, wrong schema, unmigrated) and on analytics partition runway under 2 days. Cached for 2 seconds.
Responses
{
"status": "ok",
"version": "0.1.1",
"checks": {
"database": {
"state": "ok"
},
"partitions": {
"state": "ok",
"detail": "7d"
}
},
"snapshotAgeSeconds": {
"production": 4
}
}
{
"status": "not_ready",
"version": "0.1.1",
"checks": {
"database": {
"state": "failed",
"detail": "unreachable"
},
"partitions": {
"state": "ok",
"detail": "7d"
}
}
}
Example
curl http://127.0.0.1:3000/ready
Health summary
GET/health
Auth: None.
The same body as /ready, always with 200, for a human or a status page. Wire probes to /ready.
Responses
200 OK Always.
{
"status": "ok",
"version": "0.1.1",
"checks": {
"database": {
"state": "ok"
},
"partitions": {
"state": "ok",
"detail": "7d"
}
},
"snapshotAgeSeconds": {
"production": 4
}
}
Example
curl http://127.0.0.1:3000/health
Prometheus metrics
GET/metrics
Auth: None.
Prometheus text format. The ones to alert on are in Operations.
Responses
200 OK Metrics.
# TYPE featurekit_snapshot_age_seconds gauge
featurekit_snapshot_age_seconds{environment="production"} 4
# TYPE featurekit_snapshot_compile_failures_total counter
featurekit_snapshot_compile_failures_total{reason="error"} 0
# TYPE featurekit_partition_runway_days gauge
featurekit_partition_runway_days 7
Example
curl http://127.0.0.1:3000/metrics