Health Check

Liveness only. It does not touch the database and does not call any distributor, so it stays fast and cheap enough to poll from a monitor. Note the path: it sits at the root, not under /api/v1.

curl https://api.gatewaystream.com/healthz

Response#

status

string optional

Always "ok" when the server answers at all.

service

string optional

The service name.

time

string (ISO 8601) optional

Server time when the check ran.

modes.sandbox

string optional

ready, not_configured or failed — see below.

modes.live

string optional

The same, for the live environment.

Mode States#

• ready — that environment came up and keys for it will work.
• not_configured — this deployment has no configuration for it. Keys pointing there get 503 mode_unavailable. Normal on a sandbox-only deployment.
• failed — it was configured but did not initialise. Keys pointing there will also get 503. This one is worth alerting on.

Status Codes#

200

The service is up. Read modes to see whether the environment you care about is usable.

Anything other than a 200 — including no response at all — means the service is not serving, and there is no JSON body to read.

Monitoring#

A sensible monitor polls this endpoint every thirty seconds or so and alerts on a non-200, or on the environment you depend on reading anything other than ready. Because it takes no key and touches nothing, polling it is cheap and it is exempt from the usual rate limit.

# Alert if sandbox is anything but ready
curl -s https://api.gatewaystream.com/healthz | jq -e '.modes.sandbox == "ready"'