Configuration

Every property lives under the synapsys prefix.

:::caution A typo fails startup The starter binds with ignoreUnknownFields = false. An unrecognised synapsys.* key stops the application from booting rather than being silently ignored.

This is deliberate. A misspelled heartbeat-interval-ms that binds to nothing leaves a worker running on defaults with no indication why — and that is far more expensive to find than a failed boot with the bad key named in the error. :::

Properties

Property	Type	Default	Notes
synapsys.core-url	String	—	Required. Base URL of the Core, no trailing slash.
synapsys.worker-name	String	spring.application.name	Must be stable across restarts — it is how the Core recognises a returning worker.
synapsys.host	String	machine hostname	Set it explicitly when the resolved name is meaningless, as with a container id.
synapsys.heartbeat-interval	Duration	5s	Positive whole seconds: 1s, 5s, 1m.
synapsys.fail-fast	boolean	false	Whether an unreachable Core at startup stops the application.
synapsys.connect-timeout	Duration	2s	Connect timeout for the heartbeat call.
synapsys.request-timeout	Duration	5s	Read timeout for the heartbeat call.

A full example

application.yaml

spring:
  application:
    name: billing-jobs

synapsys:
  core-url: http://synapsys-core.internal:8080
  worker-name: billing-jobs
  host: billing-01.eu-west
  heartbeat-interval: 5s
  fail-fast: false
  connect-timeout: 2s
  request-timeout: 5s

Choosing an interval

The interval is the upper bound on how long a start or stop takes to be picked up. Five seconds is a reasonable default for background jobs.

Lowering it makes the console feel more responsive and costs one HTTP request per worker per interval. Raising it beyond about thirty seconds makes the console feel broken, because a button press appears to do nothing for half a minute.

The v1 wire protocol represents this value as an integer number of seconds, so sub-second durations are rejected at startup.

fail-fast

Left false, an application whose Core is unreachable still starts, still runs its processes, and keeps retrying the heartbeat in the background. Set it true only when a worker is worthless without the Core — most are not.