Commit graph

6 commits

Author SHA1 Message Date
Oskar Kapala c255a021d1 fix(observer): quarantine malformed event files to prevent processing wedge
Was: malformed event (bad JSON / truncated / corrupted bytes) wedged the
node's checkpoint forever — every cycle re-tried, logged, never advanced
past the bad file; all subsequent good events for that node lost.

Now: first parse failure -> atomic os.replace to STATE_DIR/observer_failed_events/<node>/
with collision handling. Checkpoint advances, downstream events flow.
Move failures themselves are logged but don't crash the loop.

Complementary to yesterday's atomic_write_json fix (state files);
this addresses the same race-pattern on event files instead.

Regression test asserts: bad event quarantined to failed_events dir,
removed from hot path, subsequent good event processed (node online),
checkpoint moves to good event.
2026-06-12 11:22:56 +02:00
Oskar Kapala fa59625aa6 docs(ha-diag-agent): replace curl verify commands with docker exec
Port 8087 is no longer mapped to the host. Operator verify commands
that used curl http://localhost:8087/health now use docker exec with
Python's urllib (the image is python:3.11-slim, no curl binary).

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-06-11 19:46:33 +02:00
Oskar Kapala 52607a7cdd feat(control-plane): shadow_mode for HA event auto-actions + deploy docs
- HA_DIAG_SHADOW_MODE env flag in supervisor (default true)
- shadow_mode downgrades container_restart actions to alert_only with
  [SHADOW MODE] note; same action_id and 30-min cooldown apply
- alert_only events unaffected (always routed normally)
- 3 new tests: shadow on/off for ha_websocket_dead, alert-only unaffected
- DEPLOY.md with token gen, per-host config, verification, 48h observation,
  production-mode enablement, rollback
- README.md updated with shadow mode flag summary and DEPLOY.md link

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-29 17:12:33 +02:00
Oskar Kapala 31b48d162a feat(ha-diag-agent): WebSocketMonitor for real-time HA liveness
- persistent WS connection to HA with auth + state_changed subscription
- watchdog detects silence > 5min → emits ha_websocket_dead
- immediate ha_websocket_dead on disconnect, exponential reconnect with jitter
- cooldown prevents alert spam (10min repeat window while HA stays down)
- ha_websocket_recovered emitted on reconnect after a dead alert (allows
  supervisor to clear active incidents in Phase 5)
- new monitors/ subpackage for long-running tasks (vs interval checks/)
- /health endpoint now includes ws_connected field
- 26 unit tests, 3 integration tests (real HA + container stop/restart)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-29 15:00:18 +02:00
Oskar Kapala 07bd498fd6 feat(ha-diag-agent): test environment with dual HA Docker instances
- dockerized ken + chelsty HA test instances with template fixtures
- snapshot/reset/wait scripts for fixture management
- integration test infrastructure with separate marker
- location_tag promoted from metadata to event payload (Phase 1 flag #3)
- chelsty-infra target_url points to chelsty-ha via tailnet (Phase 1 flag #1)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-29 12:56:13 +02:00
Oskar Kapala ab8895d28b feat(ha-diag-agent): scaffold service with HA REST client and event emitter
- new per-host service, follows node-agent pattern
- 7 new HA event types defined (routing in supervisor — Phase 5)
- HeartbeatCheck as pipeline validator (pings /api/, emits ha_websocket_dead)
- service.yaml + host configs for piha (ken) and chelsty-infra (chelsty)
- test scaffolding with aiohttp/aiosqlite mocks (15/15 passing)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-29 12:26:34 +02:00