Probe behaviorRuntime safety

Runtime safety

Interpret normalized safety states, enforced limits, runtime-specific reason codes, and cooldown behavior.

Safety overview

get_safety_overview returns the broker-derived online state, latest agent state, raw detail, enforced limits, probe status counts, and caveats for each service. Green means the agent has not reported a LiveProbe safeguard trip. It is not a general application health signal.

Canonical limits

Environment variableDefaultRuntimes
LIVEPROBE_MAX_PROBE_HITS_PER_SECOND10Node, Python, JVM
LIVEPROBE_MAX_PROBE_PAUSE_MS_PER_SECOND20 msPython
LIVEPROBE_MAX_TELEMETRY_BYTES_PER_SECOND204,800Node, Python
LIVEPROBE_MAX_BUFFERED_EVENT_BYTES5 seconds of telemetry, minimum 64 KiBNode
LIVEPROBE_MAX_EVENT_LOOP_LAG_MS50 ms p95Node
LIVEPROBE_SAFETY_COOLDOWN_MS10,000 msNode, Python

Agents report only safeguards they actually enforce. An omitted limit means unsupported, not zero utilization.

Reason codes

CodeMeaning
event_loop_lagNode event-loop p95 exceeded the configured threshold
pause_budgetPython callbacks exceeded their per-second time budget
rate_limitedJVM breakpoint requests were suspended by hit-rate protection
instrumentation_failureThe runtime could not install or maintain instrumentation
agent_worker_failureAn internal agent worker failed

Runtime semantics

A red state describes LiveProbe's own runtime activity. It does not measure CPU load, memory pressure, garbage collection, or overall service availability. Reduce probe rate, remove hot-path snapshots, and wait for cooldown before retrying.