Live diagnostic actuator¶

TL;DR. /actuator/pulse (JSON) and /actuator/pulseui (zero-dep HTML) show every Pulse setting, the last OTel exporter status, and a dry-run / killswitch toggle. No redeploys to inspect a running pod.

When something looks off, the answer should never be "redeploy with debug logging." You should be able to ask the running app what it's doing, what it's configured with, and what it last sent to the Collector — without disturbing it.

Pulse exposes its full operational state through the actuator, both as JSON for tooling and as a single dependency-free HTML page for humans.

What you get¶

A single browser tab, no auth gymnastics, no extra dashboards to provision:

Endpoint	What it shows
`/actuator/pulse`	JSON snapshot of every Pulse feature and its effective configuration
`/actuator/pulseui`	Same, rendered as a single HTML page (browser-friendly)
`/actuator/pulse/runtime`	Cardinality top-offenders, SLO compliance, OTel exporter freshness
`/actuator/pulse/effective-config`	Full resolved `pulse.` configuration tree (every per-feature `Properties` record merged)
`/actuator/pulse/config-hash`	Fleet-drift hash + contributing keys
`/actuator/pulse/enforcement`	Current enforcement mode (`ENFORCING` or `DRY_RUN`); `POST` flips it at runtime
`/actuator/pulse/slo`	Generated `PrometheusRule` YAML (see SLO-as-code)

Plus the pre-built health indicators:

Endpoint	Meaning
`/actuator/health/otelExporter`	UP when the trace exporter has actually exported in the last `pulse.health.otel-exporter-stale-after` (default 5m)
`/actuator/health/dependency`	DEGRADED when caller-side error rate for any tracked downstream crosses threshold
`/actuator/health/containerMemory`	`OUT_OF_SERVICE` when headroom < `pulse.container-memory.headroom-critical-ratio`; `UNKNOWN` without cgroup accounting
`/actuator/health/jobs`	DOWN when a `@Scheduled` job hasn't succeeded inside its grace period

Turn it on¶

Nothing. The endpoints are registered automatically. Expose them through standard Spring Boot configuration:

management:
  endpoints:
    web:
      exposure:
        include: health,info,pulse,pulseui,prometheus

When to turn off the UI¶

If you don't want the HTML page exposed (e.g. on a shared management port behind a reverse proxy you don't control), use the standard Spring Boot endpoint controls — Pulse adds no proprietary toggle here:

management:
  endpoint:
    pulseui:
      enabled: false                       # disable the bean entirely
  endpoints:
    web:
      exposure:
        include: health,info,pulse,prometheus   # or simply exclude pulseui

JSON endpoints stay available either way.

Security note¶

The Pulse endpoints expose configuration and runtime state — useful for operators, sensitive in the wrong hands. Apply the same exposure controls you'd use for /actuator/env: bind management endpoints to a separate port (management.server.port) and put them behind your normal admin auth.

Source: io.github.arun0009.pulse.actuator · Status: Stable since 1.0.0