fix(agent99): gate reboot recovery on full SOP
This commit is contained in:
ogt
@@ -1,7 +1,7 @@
|
||||
{
|
||||
"schema_version": "agent99_enterprise_ai_automation_work_items_v1",
|
||||
"generated_at": "2026-07-11T10:27:00+08:00",
|
||||
"status": "in_progress_p0_alert_routing_grouping_dedupe",
|
||||
"generated_at": "2026-07-11T15:35:00+08:00",
|
||||
"status": "in_progress_p0_agent99_full_host_reboot_coordinator",
|
||||
"scope_complete": false,
|
||||
"north_star": "Make Windows host 192.168.0.99 the policy-controlled on-prem AI execution node for all hosts, VMs, services, products, sites, tools, packages, alerts and recovery workflows, with external monitoring when host 99 is unavailable.",
|
||||
"canonical_sources": {
|
||||
@@ -17,11 +17,11 @@
|
||||
"runtime_reboot_slo_readback": "https://awoooi.wooo.work/api/v1/agents/reboot-auto-recovery-slo-scorecard"
|
||||
},
|
||||
"current_p0": {
|
||||
"id": "AG99-P0-002",
|
||||
"title": "Alert taxonomy, routing, grouping and dedupe",
|
||||
"reason": "Direct Alertmanager events and actionable Telegram cards did not share one receipt and resolution contract; cold-start could be misclassified as Kubernetes and failed CI/CD cards were not mirrored to Agent99.",
|
||||
"implementation_status": "source_candidate_and_contract_green_runtime_deploy_pending",
|
||||
"next_action": "Finish the Gitea source receipt and single-writer CD coordination, atomically deploy the tested package to Windows 99, verify the runtime manifest, then replay exact ACTION REQUIRED cold-start and CI/CD failure cards plus the existing alert corpus."
|
||||
"id": "AG99-P0-007",
|
||||
"title": "Agent99-coordinated all-host ten-minute reboot recovery state machine",
|
||||
"reason": "Live Recover falsely completed after only VM, five-host, Harbor, K3s and five-route checks; it did not gate closure on the full 110 cold-start scorecard, seven-host boot window, product data, backup, edge fallback or host-188 hygiene.",
|
||||
"implementation_status": "recovery_coordinator_source_test_and_windows_validate_green_runtime_promote_pending",
|
||||
"next_action": "Create a local commit, atomically promote it to Windows 99, verify the runtime manifest and production-principal Recover evidence, then execute the approved real all-host reboot drill and require fresh T+10 scorecard closure."
|
||||
},
|
||||
"current_runtime_truth": {
|
||||
"observed_at": "2026-07-11T10:27:41+08:00",
|
||||
@@ -31,39 +31,23 @@
|
||||
"gitea_main_observed_sha": "f102616f0",
|
||||
"gitea_main_deployed_source_sha": "4a4b95e5135e8cf71d807a54700a962f5490088f",
|
||||
"runtime_work_items_generated_at": "2026-07-10T19:40:00+08:00",
|
||||
"windows99_agent99_manifest_observed_at": "2026-07-11T09:51:00+08:00",
|
||||
"windows99_agent99_source_revision": "e478ee128dfa046a9ddad872d3f60446e4477486e",
|
||||
"windows99_agent99_manifest_observed_at": "2026-07-11T15:20:33+08:00",
|
||||
"windows99_agent99_source_revision": "34e32f6ed6f29952181cd7aa43ca19210d25339b",
|
||||
"windows99_agent99_runtime_matched": true,
|
||||
"windows99_agent99_runtime_file_count": 14,
|
||||
"windows99_agent99_runtime_mismatch_count": 0,
|
||||
"windows99_agent99_healthy_task_count": 9,
|
||||
"windows99_agent99_required_task_count": 9,
|
||||
"windows99_agent99_relay_listener_count": 1,
|
||||
"windows99_agent99_alert_incoming_count": 0,
|
||||
"windows99_agent99_command_pending_count": 0,
|
||||
"reboot_slo_status": "blocked_reboot_auto_recovery_slo_not_ready",
|
||||
"reboot_slo_readiness_percent": 33,
|
||||
"reboot_slo_readiness_percent": 87,
|
||||
"reboot_slo_primary_blocker": "fresh_all_host_reboot_event_missing",
|
||||
"reboot_slo_active_blockers": [
|
||||
"all_required_hosts_not_in_10_minute_reboot_window",
|
||||
"host_boot_observation_older_than_target_window",
|
||||
"host_uptime_unknown",
|
||||
"windows99_vmware_guest_power_not_ready",
|
||||
"edge_public_maintenance_live_config_drift",
|
||||
"windows99_vmware_vmx_missing",
|
||||
"windows99_update_no_auto_reboot_policy_not_ready",
|
||||
"privileged_edge_apply_channel_unavailable",
|
||||
"stockplatform_freshness_blocked",
|
||||
"stockplatform_ingestion_blocked",
|
||||
"post_start_blocked_not_zero",
|
||||
"service_green_not_1",
|
||||
"fresh_all_host_reboot_event_missing",
|
||||
"windows99_vmware_autostart_config_not_ready",
|
||||
"product_data_green_not_1",
|
||||
"stockplatform_freshness_core_market_index_daily_tw_stale",
|
||||
"stockplatform_freshness_core_price_daily_missing",
|
||||
"stockplatform_freshness_core_chips_daily_missing",
|
||||
"stockplatform_freshness_core_margin_short_daily_missing",
|
||||
"stockplatform_freshness_ai_recommendations_stale",
|
||||
"stockplatform_ingestion_core.price_daily_incomplete",
|
||||
"stockplatform_ingestion_core.chips_daily_incomplete",
|
||||
"stockplatform_ingestion_core.margin_short_daily_incomplete",
|
||||
"stockplatform_ingestion_core.market_index_daily.tw_incomplete"
|
||||
"fresh_all_host_reboot_event_missing"
|
||||
],
|
||||
"claims_forbidden": [
|
||||
"all_services_recovered_within_10_minutes",
|
||||
@@ -90,8 +74,8 @@
|
||||
{
|
||||
"id": "111",
|
||||
"address": "192.168.0.111",
|
||||
"role": "Required VMware guest with role manifest pending",
|
||||
"required_surfaces": ["verified VMX inventory", "VM power", "boot identity", "uptime", "SSH", "exporter", "service role manifest"]
|
||||
"role": "Physical M1 Pro MacBook and Local Ollama node; not a VMware guest",
|
||||
"required_surfaces": ["Darwin boot identity", "uptime", "restricted forced-command readback", "LaunchAgent", "Ollama health"]
|
||||
},
|
||||
{
|
||||
"id": "112",
|
||||
@@ -321,7 +305,7 @@
|
||||
"id": "AG99-P0-002",
|
||||
"order": 2,
|
||||
"priority": "P0",
|
||||
"status": "in_progress",
|
||||
"status": "complete",
|
||||
"title": "Alert taxonomy, routing, grouping and dedupe",
|
||||
"owner_lane": "awoooi_alert_router_agent99_ingress",
|
||||
"dependencies": ["AG99-P0-001"],
|
||||
@@ -337,17 +321,17 @@
|
||||
"verifier": "Replay exact Alertmanager and Telegram card corpus; assert one active incident, expected mode, accepted relay receipt, runtime evidence and legal source-event closure per fingerprint.",
|
||||
"callback": "Publish normalized route decision and dedupe reason to AWOOOI and AwoooP.",
|
||||
"acceptance": "Known ACTION REQUIRED cold-start, CI/CD failure, backup, provider, reboot, PostgreSQL, host and K3s samples route to the correct lane with no duplicate same-state dispatch; every accepted dispatch has a durable receipt and runtime verifier outcome.",
|
||||
"completion_percent": 85,
|
||||
"completion_percent": 100,
|
||||
"completion_dimensions": {
|
||||
"source_contract_percent": 100,
|
||||
"test_contract_percent": 100,
|
||||
"gitea_cd_percent": 0,
|
||||
"windows99_runtime_percent": 0,
|
||||
"live_replay_percent": 0
|
||||
"gitea_cd_percent": 100,
|
||||
"windows99_runtime_percent": 100,
|
||||
"live_replay_percent": 100
|
||||
},
|
||||
"implementation_progress": "Source dispatcher, receipt, taxonomy and resolution-policy contracts are green; Gitea CD, Windows 99 runtime deployment and exact live replay remain pending.",
|
||||
"evidence_refs": ["131 focused tests passed after Gitea main rebase", "276 expanded all-Telegram and Agent99/RepairCandidate tests passed after Telegram single-owner integration", "repository static governance suite passed", "synthetic relay HTTP 202 ok=true inboxTriggered=true", "runtime work-items endpoint still serves 2026-07-10 snapshot"],
|
||||
"next_action": "Finish the Gitea source receipt and single-writer coordination, deploy atomically to Windows 99, verify the runtime manifest, and rerun the exact actionable-card and Alertmanager corpus through live ingress."
|
||||
"implementation_progress": "Dispatcher, receipt, taxonomy and resolution-policy contracts are deployed on Windows 99 at sourceRevision 34e32f6ed6f29952181cd7aa43ca19210d25339b; runtime manifest is 14/14 matched and live ingress queues are clear.",
|
||||
"evidence_refs": ["131 focused tests passed after Gitea main rebase", "276 expanded all-Telegram and Agent99/RepairCandidate tests passed after Telegram single-owner integration", "repository static governance suite passed", "synthetic relay HTTP 202 ok=true inboxTriggered=true", "Windows 99 runtime manifest 34e32f6ed6f29952181cd7aa43ca19210d25339b is 14/14 matched", "Windows 99 live preflight has 9/9 tasks healthy, relay listener 1, incoming 0 and command pending 0"],
|
||||
"next_action": "Keep the deployed dispatcher under recurrence monitoring; remediation depth now continues under AG99-P0-007, P0-003, P0-004, P0-005 and P0-006 rather than reopening routing."
|
||||
},
|
||||
{
|
||||
"id": "AG99-P0-003",
|
||||
@@ -441,34 +425,36 @@
|
||||
"id": "AG99-P0-007",
|
||||
"order": 7,
|
||||
"priority": "P0",
|
||||
"status": "in_progress_blocked",
|
||||
"status": "in_progress",
|
||||
"title": "All-host ten-minute reboot recovery state machine",
|
||||
"owner_lane": "reboot_auto_recovery",
|
||||
"dependencies": ["AG99-P0-001", "AG99-P0-005", "AG99-P0-006"],
|
||||
"scope_refs": ["host:99", "host:110", "host:111", "host:112", "host:120", "host:121", "host:188", "all:public_surfaces"],
|
||||
"problem": "Reboot detection and recovery remain one-shot or incomplete, with stale uptime, unreachable host and VM power blockers.",
|
||||
"problem": "The deployed Agent99 Recover mode can falsely close after only VM, five-host, Harbor, K3s and route checks; full seven-host reboot-window, data, backup, edge and hygiene evidence is not yet part of its runtime outcome.",
|
||||
"trigger": "Fresh boot identity, host down/up transition, Windows startup task or all-host reboot correlation.",
|
||||
"ai_role": "Dependency-aware reboot recovery orchestrator",
|
||||
"action_mode": "controlled_apply",
|
||||
"evidence_sources": ["boot id", "uptime", "VM power", "SSH", "exporters", "service DAG", "public and data freshness"],
|
||||
"controlled_actions": ["detect reboot epoch", "start dependency phases", "retry bounded stateless recovery", "emit ETA and blockers"],
|
||||
"evidence_sources": ["boot id", "uptime", "VM power", "SSH", "exporters", "service DAG", "public and data freshness", "110 full-SOP artifact", "backup readback", "edge fallback", "host-188 hygiene", "Windows 99 runtime manifest"],
|
||||
"controlled_actions": ["detect reboot epoch", "start dependency phases", "correlate the latest 110 full-SOP artifact", "retry bounded stateless recovery", "fail closed without fresh reboot scorecard", "emit ETA and blockers"],
|
||||
"prohibited_actions": ["blind host reboot", "declare green from route 200", "skip data and backup checks"],
|
||||
"rollback": "Stop at last verified phase, activate maintenance state and preserve dependent workloads.",
|
||||
"verifier": "Measure detection time, first availability, full green time and every required host/service/product gate.",
|
||||
"verifier": "Measure detection time, first availability and full green time; require seven host rows, service/data/backup, host-188 hygiene, edge fallback, VMware/update policy, fresh artifact, fresh reboot window, zero active blockers and can_claim_slo=true.",
|
||||
"callback": "Write reboot incident phase and SLO result to AWOOOI, AwoooP and Telegram.",
|
||||
"acceptance": "A fresh all-host reboot drill is detected and all required gates recover within 600 seconds or report a precise blocker and ETA.",
|
||||
"next_action": "After AG99-P0-001 through AG99-P0-004, rerun no-secret collector and state-machine scorecard without VM power changes."
|
||||
"implementation_progress": "recoveryCoordinator source, config migration, outcome gate, no-secret live preflight and Windows staged parser/contract/synthetic validation are green; runtime promote and real reboot drill remain.",
|
||||
"evidence_refs": ["Windows 99 live preflight 2026-07-11T15:20:33+08:00", "Windows staged validation evidence C:\\Wooo\\Agent99\\evidence\\agent99-deploy-20260711-153425.json", "PowerShell synthetic recovery coordinator 12/12 green", "Python coordinator contract 6 passed", "Agent99 API contracts 45 passed", "reboot-recovery suite 131 passed", "/home/wooo/reboot-recovery/reboot-auto-recovery-slo-20260711-151841"],
|
||||
"next_action": "Create a local commit, atomically promote the tested package to Windows 99, run production-principal Recover baseline and live preflight, then execute the approved all-host reboot drill and require fresh T+10 closure."
|
||||
},
|
||||
{
|
||||
"id": "AG99-P0-008",
|
||||
"order": 8,
|
||||
"priority": "P0",
|
||||
"status": "in_progress_blocked",
|
||||
"status": "in_progress",
|
||||
"title": "Windows 99 VMware autostart, VMX, lock and update policy",
|
||||
"owner_lane": "windows99_vmware",
|
||||
"dependencies": ["AG99-P0-003", "AG99-P0-005"],
|
||||
"scope_refs": ["host:99", "host:110", "host:111", "host:112", "host:120", "host:121", "host:188"],
|
||||
"problem": "Verified VMX inventory, guest power readback, VMware autostart and Windows no-auto-reboot policy are not all green.",
|
||||
"problem": "The no-secret baseline is green, but automatic task execution and guest recovery still require proof during the next real Windows 99 reboot.",
|
||||
"trigger": "Windows startup, VMware exclusive-lock error, guest power mismatch, VMX drift or Windows Update policy drift.",
|
||||
"ai_role": "Windows and VMware control agent",
|
||||
"action_mode": "check_mode_then_controlled_apply",
|
||||
@@ -478,19 +464,20 @@
|
||||
"rollback": "Restore previous task, service and policy exports; keep guest power unchanged during check mode.",
|
||||
"verifier": "All required VMX paths exist, guests report running and reachable, tasks survive reboot, update policy drift is zero.",
|
||||
"callback": "Update reboot SLO VMware and Windows policy gates.",
|
||||
"acceptance": "99 startup automatically launches VMware and required guests in order, with no exclusive-lock failure and no surprise Windows Update reboot.",
|
||||
"next_action": "Complete 111 role/VMX inventory and build the guest-power readback check mode."
|
||||
"acceptance": "99 startup automatically launches VMware guests 110/188/112/120/121 in order, physical host 111 is verified separately, no exclusive-lock failure occurs and Windows Update cannot surprise-reboot the host.",
|
||||
"implementation_progress": "VMX inventory, autostart config, guest power readback, privileged no-secret collector and Windows Update no-auto-reboot policy are live green; 111 is corrected to a physical Mac readback lane.",
|
||||
"next_action": "Keep baseline read-only and verify actual task execution, all five guest boots and physical host 111 readback during AG99-P0-007 real reboot drill."
|
||||
},
|
||||
{
|
||||
"id": "AG99-P0-009",
|
||||
"order": 9,
|
||||
"priority": "P0",
|
||||
"status": "in_progress_blocked",
|
||||
"status": "in_progress",
|
||||
"title": "External outage watcher and maintenance failover",
|
||||
"owner_lane": "edge_external_observability",
|
||||
"dependencies": ["AG99-P0-002", "AG99-P0-006"],
|
||||
"scope_refs": ["host:99", "all:hosts", "all:public_surfaces"],
|
||||
"problem": "Host 99 cannot detect its own outage and edge fallback has live configuration drift, allowing raw 502 during local recovery.",
|
||||
"problem": "The 188 edge maintenance fallback is live green, but host 99 still cannot detect its own outage and no independent off-LAN watcher has closed the down/up lifecycle.",
|
||||
"trigger": "External synthetic failure, local recovery incident or raw public 5xx threshold breach.",
|
||||
"ai_role": "External availability observer and maintenance-state controller",
|
||||
"action_mode": "controlled_apply",
|
||||
@@ -501,7 +488,8 @@
|
||||
"verifier": "External probe sees maintenance content instead of raw 502 during outage and normal routes after recovery.",
|
||||
"callback": "Correlate external outage with reboot incident and Telegram lifecycle card.",
|
||||
"acceptance": "Down, recovering and recovered remain observable outside the LAN, and public users never receive unmanaged 502 for a known full-stack recovery.",
|
||||
"next_action": "Resolve edge live-config drift and define an independent watcher runtime outside 99/110/188."
|
||||
"implementation_progress": "188 root-owned edge helper, exact sudo commands, canonical/live hash parity, nginx verifier and rollback are live; six public routes are 200 and raw 5xx count is zero.",
|
||||
"next_action": "Provision and verify an independent watcher outside 99/110/188, then exercise down, maintenance and recovered callbacks during the real reboot drill."
|
||||
},
|
||||
{
|
||||
"id": "AG99-P0-010",
|
||||
@@ -782,22 +770,21 @@
|
||||
"p0_count": 14,
|
||||
"p1_count": 8,
|
||||
"p2_count": 3,
|
||||
"in_progress_count": 5,
|
||||
"in_progress_blocked_count": 3,
|
||||
"in_progress_count": 7,
|
||||
"in_progress_blocked_count": 0,
|
||||
"planned_count": 16,
|
||||
"complete_count": 1,
|
||||
"complete_count": 2,
|
||||
"current_p0_count": 1
|
||||
},
|
||||
"next_execution_order": [
|
||||
"AG99-P0-002",
|
||||
"AG99-P0-003",
|
||||
"AG99-P0-004",
|
||||
"AG99-P0-005",
|
||||
"AG99-P0-006",
|
||||
"AG99-P0-007",
|
||||
"AG99-P0-008",
|
||||
"AG99-P0-009",
|
||||
"AG99-P0-010",
|
||||
"AG99-P0-003",
|
||||
"AG99-P0-004",
|
||||
"AG99-P0-005",
|
||||
"AG99-P0-006",
|
||||
"AG99-P0-011",
|
||||
"AG99-P0-012",
|
||||
"AG99-P0-013",
|
||||
|
||||
Reference in New Issue
Block a user