User Guide Monitoring

User Guide - Monitoring

This page covers how to run anomaly monitoring workflows for chargers.

When To Use This Page

Use this page when starting, validating, or stopping monitoring workloads from the UI.

Audience

• Operators managing anomaly monitoring workloads.

What Monitoring Does

• Creates a RADAR workload for selected MQTT topics.
• Applies selected model and optional preprocessors.
• Writes anomaly records to platform database when DB write mode is enabled.

Standard Workflow

1. Open /monitoring/:chargerId.
2. Select telemetry topic(s) for the charger.
3. Select model type.
4. Optionally tune model parameters and preprocessors.
5. Start monitoring.
6. Observe workload status and anomaly output.

Model And Preprocessor Selection Guidance

• Start with defaults for first-run validation.
• Tune model parameters only after baseline signal quality is confirmed.
• Add preprocessors when telemetry scale or feature shape needs normalization.

Stop / Cleanup Workflow

1. Open monitoring view for charger.
2. Stop active workload by container_name or service reference.
3. Confirm state is stopped/removed in monitor list.

If lifecycle is ephemeral, workloads are cleaned automatically when TACTIC restarts.

Operational Checks

curl http://localhost:8000/v1/monitors/all
curl "http://localhost:8000/v1/monitors?container_name=<name>"
curl http://localhost:8000/v1/anomalies/count

Failure Modes

Symptom	Likely Cause	Action
Start fails with validation error	invalid model/preprocessor params	revert to defaults, retry, then incrementally tune
Service runs but no anomalies appear	no data flow or warmup not reached	verify MQTT flow, wait for baseline window
Service disappears after restart	ephemeral lifecycle configured	set TACTIC_RADAR_WORKLOAD_LIFECYCLE=persistent if required
Models/preprocessors endpoint empty	model registry not initialized	check TACTIC /ready and middleware logs

Related Pages

• Backend API Guide
• Architecture Overview
• Environment Variables
• Testing & Debugging