Fixes interface auto-detection failures on Pi Zero 2W where wlan0 doesn't exist
at boot. Implementation:
- Added detect_interface_with_retry() function with 3x retry fallback chain
- Config-supplied interface preferred, then /proc/net/route, then UP interface
- Exponential backoff: 5s, 10s, 20s delays between retries
- Filters excluded prefixes: lo, tailscale*, wg*, tun*, docker*, veth*, br-*
- Used in Engine.__init__(), PacketCapture.start(), MacManager.start()
- Full test coverage: 10 tests for detection, backoff, filtering
Addresses issue where network stack initializes after boot and interfaces
become available after 5-10 second delay.
build_data.py created 'profiles' but mac_manager queries 'mac_profiles'.
Columns category/model renamed to device_type/device_name. device_type
values aligned with _CATEGORY_MAP in mac_manager (streaming, smart_tv,
phone, tablet, smart_speaker, iot, printer, gaming).
ja3_fingerprints.db table renamed from 'fingerprints' to 'ja3_profiles'
and expanded with all columns ja3_spoofer queries (name, description,
cipher_suites, extensions, elliptic_curves, ec_point_formats as JSON).
StateManager.reset_module_status() clears stale 'running' entries on
startup so watchdog doesn't flag old-session PIDs as dead modules.
Engine.start_all() calls reset before launch and does a 3s post-startup
liveness check, logging any modules that crashed immediately after fork.
With multiprocessing fork, threads are not inherited. The parent's
CaptureBus reader thread runs in the parent but module subprocesses
subscribe AFTER fork, so their SubscriberQueue entries are added to
a child-local copy of _subscribers that the parent reader never sees.
Fix: reset and restart CaptureBus in each module subprocess so it
has its own reader thread on the inherited AF_PACKET socket. Each
module gets independent packet delivery. Linux AF_PACKET allows
multiple readers on the same interface.
Three bugs causing all modules to exit within seconds of startup:
1. _module_runner had no keep-alive loop after calling module.start().
All modules use a thread-spawning pattern where start() returns
immediately, so the subprocess exited and killed all daemon threads.
Added while module._running: sleep(1) loop to block until stop().
2. Engine injected capture_bus under key 'capture_bus' but 9 modules
(auth_flow_tracker, cloud_token_harvester, ldap_harvester, network_mapper,
rdp_monitor, quic_analyzer, smb_monitor, db_interceptor, vlan_discovery)
look for '_capture_bus'. Inject both keys so all modules get it.
3. mac_manager crashes on startup when innocuous_macs.db exists but has
no mac_profiles table (DB not yet seeded). Added OperationalError
handler to fall back to random OUI instead of crashing.
- Engine.__init__ now initializes self.capture_bus = None
- Engine.start_all() instantiates CaptureBus before starting modules
- CaptureBus uses resolved interface from config
- Inject capture_bus into configs of all modules with requires_capture_bus=True
- Engine.stop_all() properly stops CaptureBus after stopping modules
- CaptureBus instantiation includes error handling with fallback
- Engine.__init__ now detects actual interface when config has 'auto'
- Uses get_primary_interface() to find default route interface
- PacketCapture module updated to use detected interface with fallback
- Handles Pi Zero 2W (wlan0) and other non-eth0 systems
- Add pi3b tier (700MB RAM limit, 8 passive/2 active modules, no bettercap/mitmproxy)
- Update detect_hardware_tier() to detect Raspberry Pi 3B by model string and BCM2837
- Add pi3b section to hardware_tiers.yaml
- Add systemd MemoryMax/CPUQuota/OOMScoreAdj to all three service files
- Fix bigbrother-watchdog.service ExecStart (was calling non-existent ResourceMonitor().run())
- Add scripts/run_watchdog.py as standalone health monitor daemon
Cobra