Troubleshooting

Step 1 — Verify USB enumeration

lsusb | grep -iE "analog|pluto|hamgeek"
dmesg | grep -i pluto

If the device does not appear in lsusb output, try a different USB port or cable. USB 3.0 ports are required for full-bandwidth operation.

Step 2 — Check IIO subsystem

iio_info -s

This should list all available IIO contexts. For network-attached PlutoSDR units:

iio_info -u ip:192.168.2.1

Step 3 — Verify network link (network-mode SDRs)

ip link show
ping -c 3 192.168.2.1

Step 4 — Driver check

modinfo industrialio
dmesg | tail -30

Step 1 — Test broker reachability

mosquitto_sub -h broker.local -p 8883 -t '#' --cafile /etc/varta/certs/ca.pem -v

If connection is refused, verify the broker process is running:

systemctl status mosquitto
journalctl -u mosquitto --since "10 min ago"

Step 2 — Check certificate expiry

openssl x509 -in /etc/varta/certs/client.pem -noout -dates
openssl x509 -in /etc/varta/certs/ca.pem -noout -dates

Step 3 — Verify mTLS handshake

openssl s_client -connect broker.local:8883 \
  -CAfile /etc/varta/certs/ca.pem \
  -cert /etc/varta/certs/client.pem \
  -key /etc/varta/certs/client.key

Step 4 — Inspect broker ACLs

Ensure the sensor's client ID has publish/subscribe permissions on the required topics (varta/sensors/+/detections, varta/health/+).

Step 1 — Check baseline calibration

curl -s http://localhost:8080/api/v1/sensor/baseline | python3 -m json.tool

If baseline_age_seconds exceeds 86400 (24 hours), recalibrate:

curl -X POST http://localhost:8080/api/v1/sensor/recalibrate

Step 2 — Verify CNN model version

curl -s http://localhost:8080/api/v1/model/info

Compare the model_version field against the latest release. Outdated models miss newer drone protocols.

Step 3 — Check signature database

ls -la /opt/varta/data/signatures/
cat /opt/varta/data/signatures/version.json

Signature database updates are distributed via the Varta update channel. Ensure the node has pulled the latest definitions.

Step 4 — Review detection thresholds

In /etc/varta/detection.yaml, verify that confidence_threshold and snr_minimum are appropriate for the deployment environment. Urban environments typically require higher thresholds (0.75+) to suppress noise.

Step 1 — Query the health API

curl -s http://localhost:8080/api/v1/health | python3 -m json.tool

Key fields to inspect: state, uptime_seconds, last_calibration, error_count.

Step 2 — Check for stale baseline

A baseline older than 24 hours in a changing RF environment will cause drift. The health API reports baseline_stale: true when this condition is detected.

Step 3 — Restart the sensor service

sudo systemctl restart varta-sensor
journalctl -u varta-sensor -f

Watch for initialization errors during startup. A clean start should report Sensor initialized, state=NOMINAL within 30 seconds.

Step 4 — Force recalibration

curl -X POST http://localhost:8080/api/v1/sensor/recalibrate \
  -H "Content-Type: application/json" \
  -d '{"force": true, "duration_seconds": 120}'

Step 1 — List audio capture devices

arecord -l

Verify the expected microphone array appears in the device list. If missing, check USB connections and ALSA configuration.

Step 2 — Test audio capture

arecord -D hw:0,0 -f S16_LE -r 48000 -c 1 -d 5 /tmp/test_capture.wav
aplay /tmp/test_capture.wav

Verify the recording contains actual audio, not silence or noise artifacts.

Step 3 — Verify sample rate configuration

The acoustic model expects 48 kHz input. Check /etc/varta/acoustic.yaml:

cat /etc/varta/acoustic.yaml | grep sample_rate

Mismatched sample rates cause the classifier to produce garbage outputs.

Step 4 — Check model files

ls -la /opt/varta/models/acoustic/
md5sum /opt/varta/models/acoustic/classifier_v*.onnx

Compare checksums against the release manifest to rule out corrupted model files.

Step 5 — Review pipeline logs

journalctl -u varta-acoustic --since "30 min ago" --no-pager

Step 1 — Check chrony status

chronyc sources -v
chronyc tracking

The System time offset should be under 1 ms for TDOA eligibility. If offset exceeds 10 ms, investigate the time source.

Step 2 — Verify GPS daemon (if GPS-disciplined)

systemctl status gpsd
gpspipe -w -n 5

Confirm the GPS receiver has a 3D fix. Indoor installations may need an external antenna.

Step 3 — Check PTP status (if using PTP)

pmc -u -b 0 'GET CURRENT_DATA_SET'
journalctl -u ptp4l --since "10 min ago"

Step 4 — Query the time sync API

curl -s http://localhost:8080/api/v1/time/status | python3 -m json.tool

Fields: sync_source, offset_us, tdoa_eligible, last_sync.

Step 5 — Force NTP resync

sudo chronyc makestep
chronyc sources

Step 1 — Check system resource usage

htop
free -h
df -h /opt/varta

Step 2 — Jetson-specific GPU monitoring

tegrastats --interval 1000

Monitor GPU utilization, memory bandwidth, and thermal throttling. If the GPU temperature exceeds 85°C, check cooling and airflow.

Step 3 — Check inference latency

curl -s http://localhost:8080/api/v1/metrics | grep inference_latency

The p99_latency_ms value should remain below 200 ms for Varta Pro and 400 ms for Varta Mini.

Step 4 — Identify memory pressure

journalctl -k | grep -i "oom\|out of memory"
cat /proc/meminfo | head -5

Step 5 — Model optimization check

Ensure TensorRT-optimized models are deployed on Jetson platforms. Non-optimized ONNX models run 3–5x slower:

ls -la /opt/varta/models/*.trt
cat /opt/varta/config/inference.yaml | grep engine_type

System journal (last 2 hours)

journalctl --since "2 hours ago" --no-pager > /tmp/varta-journal.log

Application logs

cp /var/log/varta/*.log /tmp/varta-app-logs/
tar czf /tmp/varta-app-logs.tar.gz /tmp/varta-app-logs/

Sensor health snapshot

curl -s http://localhost:8080/api/v1/health > /tmp/varta-health.json
curl -s http://localhost:8080/api/v1/metrics > /tmp/varta-metrics.json

I/Q capture export (if RF issue)

curl -X POST http://localhost:8080/api/v1/capture/export \
  -H "Content-Type: application/json" \
  -d '{"duration_seconds": 10, "output": "/tmp/iq_capture.bin"}'

System info

uname -a > /tmp/varta-sysinfo.txt
cat /etc/varta/version >> /tmp/varta-sysinfo.txt
lsusb >> /tmp/varta-sysinfo.txt
ip addr >> /tmp/varta-sysinfo.txt

Bundle everything

tar czf /tmp/varta-diagnostics-$(hostname)-$(date +%Y%m%d).tar.gz \
  /tmp/varta-journal.log \
  /tmp/varta-app-logs.tar.gz \
  /tmp/varta-health.json \
  /tmp/varta-metrics.json \
  /tmp/varta-sysinfo.txt