Spanda

Crash Detection

Spanda detects process crashes, runtime panics, provider crashes, package crashes, sensor/actuator driver crashes, network stack crashes, and abnormal shutdowns.

Detection

Crash detection triggers on:

• Abnormal process exit codes
• Unhandled runtime panics
• Provider or package termination
• Sensor or actuator driver failure
• Network stack crash

Configuration

restart_policy ProviderRuntime {
    max_restarts: 3 within 5 min;
    on_exceeded {
        enter degraded_mode;
        disable provider;
        request_operator_review;
    }
}

on runtime crash {
    diagnose root_cause;
    recover using RestartRuntime;
}

CLI

spanda fault scan rover.sd --inject-crash
spanda fault report rover.sd --json

Recovery

On crash, Spanda can:

1. Diagnose root cause from trace and static analysis
2. Attempt recovery via configured recovery_policy
3. Record crash event in mission trace for replay
4. Lower readiness score and block deployment if critical

Recovery actions never bypass safety validation.

See also

• Runtime Fault Detection
• Recovery policies

This site is open source. Improve this page.