FAQ:
Fault Decision Matrix | Symptom | Read-Only Inspection & Diagnosis | Controlled Recovery Procedure | Strictly Prohibited Actions (Stop Conditions) |
|---|---|---|---|
| Connection Failure | Check LAN, local subnet, ping, REST 9091, SSH 22 | Fix dev PC NIC configuration or use a known healthy connection path | Do NOT reset router, change robot static IP, or restore factory settings |
| Service List active but Node Info shows no node | Source both ROS 2 environments; run ros2 daemon stop and re-check; match exact process names | Rebuild ROS 2 daemon; restart recorder ONLY when no process is running, no active recording exists, and authorization is given | Do NOT proceed if same-named processes exist, recording is active, or software version is unknown |
| Topic Data Missing | Run topic type/info -v/echo --once; verify publisher, QoS profile, and Domain ID | Fix message package dependencies, QoS settings, or subscription configurations | Do NOT switch operating modes, reboot system, or stop unknown services due to single Topic timeout |
| Control Command Ineffective (No Motion) | Check control authority, mode, E-stop, enable state, error bits, joint limits, units, and readbacks | Execute verified stop command; capture baseline state before recovery | Do NOT continuously resend commands, force-enable, clear error bits, or alter joint limits upon initial success response |
| Recording Failure / Timeout | Verify 6 required fields and integer operate parameter; check Service, exact process, recording/completed directories | On "Unknown Outcome", execute stop command first and query unique run ID | Do NOT blindly retry start commands, spawn secondary instances, or clear root directories |
| Slow System Service Restart / Startup Error Logs | Check systemctl show/status, MainPID, Result, and journal logs for exact time window; re-verify version, Topic/Service, error bits, zero velocity, cameras, and recording services | Allow a SINGLE authorized start recovery attempt; record exit timeouts and non-TTY commands for unit service rectification | Do NOT repeatedly restart services, force-terminate processes using fuzzy matching, or declare full stack failure/success based on single error logs |
| Camera Frame Rate Fluctuation | Measure one channel at a time; check camera FPS status array and MCAP frame count/duration; compare camera_info vs. image topics | Evaluate reducing stream count/resolution or staggering processing workloads | Do NOT use multi-channel parallel topic hz commands for root cause assignment or apply global tuning blindly |
| Camera FPS Array Reads 0, but H.264 Streams Active | Access target .../h264 stream using BEST_EFFORT QoS to read headers; check encoding log frame counts, queue depth, frame drops, and latency | Confirm monitoring array mappings first; validate based on continuous decoding results of target stream | Do NOT reboot cameras due to zero-value FPS array elements or declare the stream as working properly based on isolated frame timestamps |
| Aggregated Health Score 100, but Log Reports Component Disconnection | Cross-verify current hardware variant, target endpoints, /diagnostics topic, joint error bits, and log entries within exact time window | Document configuration if optional component is not applicable; escalate to support if required component is unreachable | Do NOT rely on aggregated health scores as sole judgment criteria, clear errors blindly, or continuously ping unknown devices |
| Data Tracing Heartbeat Frequency Misaligned with Definition | Inspect TraceHeartbeat interface, node startup logs, current configuration, and observe over a window exceeding 35 seconds | Interface comments and configurations are subject to unified corrections in later releases. Do not use this document until those updates have been applied. | Do NOT treat /trace/heartbeat/status as an availability criterion or invoke write-configuration interfaces prior to fix |
| Recording Directory Exists but NVMe Not Mounted | Cross-verify using findmnt -T /home/realman/ssd, df -h, block device origins, and directory disk usage levels | Stop new recording sessions immediately; request administrator to restore correct mount point | Do NOT mistake empty mount point directories on root partition for SSD storage or allow data logging to fall back to system disk |
| Incomplete TF Tree | Check /tf, /tf_static, frame connectivity, and TRANSIENT_LOCAL durability policy | Reconstruct TF using URDF + joint angles for arms/head/waist, and extrinsic calibration parameters for cameras | Do NOT forge static transforms to mask broken transform trees |
| HTTP Bridge 401 / 502 / Timeout | 401: Inspect token injection without printing plain text; 502/Timeout: Verify ROS 2 state and actual recording status | Restore credentials, SSH tunnels, or bridge services while maintaining loopback binding | Do NOT rebind service to 0.0.0.0 to bypass faults or blindly retry start commands |
| Insufficient Disk Space | Execute df, run du on exact directories, and verify run ID, fetch, parsing, and checksum results | Purge EXACT confirmed and authorized subdirectories ONLY | Strictly PROHIBIT wildcards, prefix matching, root directory operations, or "delete-upon-fetch" execution |