Skip to main content

Deployment & Operations

This guide covers the complete lifecycle of an NE503 deployment, from first installation and handover to long-term maintenance: pre-deployment checks, network and time configuration, log and disk management, and firmware upgrade and recovery. All operations are performed through the Web console's Settings and Maintenance pages or via SSH. For first-time connectivity and a basic walkthrough, see the Quick Start.

1. Pre-Deployment Checklist

Before handover, verify each item below in order. Do not proceed to the next item until the current one passes.

#ItemPass criteria
1Power & mountingPoE (802.3at) or DC 12V supply working (~5–6 W); device boots automatically on power-up
2NetworkCable connected; device IP reachable via ping
3Web console reachablehttps://<device-ip> loads the login page (accept the self-signed certificate on first access)
4Default passwords changedBoth the Web console (admin / password) and SSH (root / root) passwords changed
5Time & timezoneTimezone matches the site; NTP sync succeeds; on-stream timestamp is correct
6Streams workingRTSP main/sub streams playable: ffprobe -rtsp_transport tcp rtsp://<device-ip>:8554/main
7AI services readyPreloaded models healthy on Models; demo app starts normally
8PeripheralsAlarm / Wiegand / audio wired as needed and enabled on Peripherals
9Security baselineDevice not exposed to the internet; only necessary ports allowed — see Security Hardening
10Handover recordDevice IP / MAC, firmware version, location, and owner recorded

Configuration methods for each item are in the sections below.

2. Network and Time Configuration

Two foundational configurations must be completed during deployment: a fixed device IP (an IP change breaks NVR / platform integrations) and time synchronization (wrong timezone or clock misaligns log and event timestamps, hurting diagnosis and reconciliation).

2.1 Static IP Configuration

Go to Settings → Network, IPv4 Settings (interface eth0):

Field descriptions:

FieldDescription
ModeDHCP (router-assigned) or Static Address (manual, fixed)
IP Address / Subnet Mask / GatewayRequired in static mode
DNS Server / Secondary DNSDNS servers

Procedure:

  1. Switch Mode to Static Address.
  2. Fill in the IP address, subnet mask, and gateway per the site network plan; set DNS as appropriate for the environment.
  3. After saving, the device serves at the new IP; access the console via the new address.

Completion criteria: the IP persists across reboots; the Web console is reliably reachable at that IP; downstream systems (NVR / business platform) have been updated with the new address.

In a DHCP environment where static configuration is not possible, bind the device IP to its MAC address on the router to achieve the same fixed result.

2.2 Timezone and NTP Configuration

Go to Settings → Time:

Field descriptions:

FieldDescription
TimezoneSite timezone
Time Format12 / 24 hour
Sync ModeNTP Sync (network time) or Manual Setup
NTP ServerNTP server address (e.g. pool.ntp.org)
Sync IntervalSync interval

Procedure:

  1. Select the site timezone under Timezone.
  2. Set Sync Mode to NTP Sync and enter an NTP server reachable from the site.
  3. Click Sync Now to synchronize immediately.

Completion criteria: Sync Now completes without error; the on-stream timestamp matches local standard time.

3. Logs, Storage, and Maintenance Tools

3.1 Log Viewing and Collection

Daily inspection is done in Maintenance → Logs, which offers three views:

ViewContent
Operation LogsLogins, app start/stop, configuration changes
System LogsSystem-level events
Developer LogsDeveloper debug logs

Filter by time range and keyword; each entry has Detail for full information.

For remote collection and deep diagnosis:

ChannelMethodUse case
REST APIGET /api/v1/apps/<id>/logs (NDJSON stream)Remote app-log collection
SSHjournalctl -u <service>; platform log files under /data/aipc/logs/Deep platform-service diagnosis

Common service names for journalctl: app-manager (app management), ai-runtime (AI inference), camera-daemon (camera), event-bus (event bus).

3.2 Storage and Disk-Space Planning

Settings → Storage shows built-in storage usage, split into System (system partition, not modifiable) and Data (data partition).

Space usage reference (measured on a 2026-08 firmware sample device):

Partition / directorySizePurposePlanning
Root /3.3 G (system uses ~1.8 G)System, read-onlyNot writable, no planning needed
/data54 GData partitionBudget for the items below
/data/aipc/models20 M–3 G eachModel libraryDetection models ~20 M, VLM ~3 G; estimate by planned imports
/data/aipc/images & apps~100–450 M per appApp images and instancesEstimate by app count
/data/aipc/logsCan exceed 1 GPlatform logsClean regularly

Storage management points:

  • Alert threshold: when Data usage exceeds 80%, clean up logs, uninstall idle apps, or remove unused models; a microSD card can be inserted to expand capacity.
  • Log cleanup: via SSH, run truncate -s 0 /data/aipc/logs/*.log.
  • Video recording: the current firmware does not store video locally; recording is done by an external NVR / VMS pulling the streams (see Video and Imaging).
  • Command-line management: use aipc-cli; common commands in the Device Management Tools.

3.3 Maintenance Tools

The Maintenance page provides three more tools:

File Manager: browse and operate the device's /data/aipc filesystem. Supports path navigation and Download / Delete / Upload File / New Folder operations; each file supports Preview / Delete / More.

Terminal: a built-in Web SSH terminal providing command-line access without installing a client. Ready when the status shows Connected.

Process Manager: lists all running processes (PID / CPU / memory / status, etc.), supports filtered search, and allows Kill on abnormal processes.

Killing platform-critical processes (camera-daemon, ai-runtime, platform-api, event-bus) breaks functionality until a reboot; avoid unless necessary.

4. Firmware Upgrade and Recovery

4.1 Two-Layer Upgrade Mechanism

NE503 has two independently upgradable layers:

TrackScope
Firmware UpdatePlatform services, HAL, Web console (application layer)
System OS UpdateLinux kernel, device tree, rootfs (OS layer)

Go to Settings → Device Info; both current versions appear under Firmware & Hardware, each with an Update button. Device identity (model, serial number, MAC, IP) and uptime are also shown on this page.

Upgrade procedure: click Update beside the version → upload the package → the system parses, validates, and writes it → the device reboots automatically on completion.

Note: an upgrade takes ~2–5 minutes. Never power off or operate the device during this window; schedule upgrades for off-peak hours.

4.2 Upgrade Compatibility Validation

Before installation, every package passes six compatibility gates (machine, product, hardware compatibility, compat level, data schema, minimum recovery version). Any failing gate rejects the package automatically — nothing is written to the device. Gate details in the Version Matrix.

4.3 Rollback and Recovery

ScenarioMethod
Platform services need rollbackVia SSH, run deploy.sh --rollback (shipped with the release package) to roll back to the previous version; --status shows current state
System upgrade failedThe device uses A/B double-copy plus recovery mode and can boot from the recovery partition; see os-upgrade.md in the open-source repo
Full reset neededThe current firmware has no one-click factory reset; reflash the device per System Flashing. Flashing wipes all /data content (models, apps, logs) — back up first
Interface-board MCU upgradeThe MCU firmware upgrades separately via OTA (ne503_ota_package_v<X.Y.Z>.bin); see baseboard-mcu-rtc-ota.md

5. Production Security Baseline

Two security actions are mandatory before handover:

  • Change default credentials: the Web console and SSH default passwords (admin / password, root / root) must be changed before handover; see Security Hardening for the procedure.
  • Minimum exposure: keep the device off the public internet; allow only necessary ports and source IPs; disable SSH outside maintenance windows.

For full port lockdown, credential hardening, and least-privilege configuration, see Security Hardening.