Accepts a ZipkinConfig or plain options.
The trace query client (shares this client's connections).
The span reporter (shares this client's connections; no compression, 512 spans per POST).
Shuts down the tracing handles created through this client (flushing their batch processors), then releases connections. Idempotent.
A reporter with other options that shares this client's connections.
Liveness: GET /health → {"status":"UP"}. A 503 (DOWN, storage unavailable) gives
healthy: false without retries. Never throws for an unhealthy or unreachable service (only
CANCELLED and CONFIG).
Starts OpenTelemetry tracing to this Zipkin (see the package-level initTracing). The
native exporter uses this client's endpoint, headers, TLS, proxy, retries, hooks and logger;
an explicit config.endpoint wins. The handle is shut down by close. On a closed
client it throws ConfigError (CONFIG): starting tracing is a local setup step, not a
request, so it does not fail with CANCELLED like the calls.
Round trip through the query API and storage: GET /api/v2/services (through a gateway this
also proves the auth headers). Returns the latency in ms.
Build information: GET /info → {"zipkin":{"version","commit"}}.
Server version (e.g. 3.6.1) from GET /info. Throws UnsupportedError when the server
(or a gateway) does not report one.
StaticfromBuilds a client from GROW_ZIPKIN_* (and vendor) environment variables.
Client for Zipkin (port 9411): health, ping and server version, plus the query API (query), the direct span reporter (reporter) and tracing setup (initTracing). close also shuts down every tracing handle created through this client.