docs/RESUME.md

Session Resume

Last updated: 2026-05-23 (session continued ×4).

Open items (urgency order)

🔴 Critical — security risk or unrecoverable data loss

• Password rotation: tapirnase — shared across WiFi PSK, LiteLLM root key (sk-tapirnase), D-Link admin, Backrest repo password. Single sniff = broad blast radius. Rotate per-service. See services/homelab-architecture.md §operational rules.
• UNAS personal data off-host — ~820 GB Immich library partially covered by media-backup-plan in Backrest (last OK 2026-05-21 but failing/orphaned). video-projects-plan never completed. Backrest is the right vehicle — fix reliability first (see 🟠 Backrest item), then verify JottaCloud coverage. See services/backrest.md phase 1a.
• Nextcloud Borg passphrase → Vaultwarden — passphrase only in container env; CT 105 loss = unrecoverable backup. Move to Vaultwarden. See services/backrest.md blind spot #4b.

🟠 High — known broken / verification needed

• Arcane OIDC secret — fixed 2026-05-23. New secret generated, Pocket-ID bcrypt updated, Arcane compose updated, force-recreated. Login confirmed working.
• Backrest media-repo — video-projects-plan has NEVER completed a successful snapshot. media-backup-plan last OK 2026-05-21; failed 2026-05-22 01:30; orphaned restic run in progress since 19:10 (Backrest restarted 21:23, orphaned it). Both plans next scheduled 2026-05-25 02:00/01:30. Monitor for completion; if they fail again, investigate rclone:jottacloud:media auth.
• AdGuard split-horizon DNS — internal clients still resolve to public IPs. Rewrite rules audit. See services/adguard-dns.md.
• WAL-G monitoring — done 2026-05-23. Textfile collector (/usr/local/bin/walg-metrics.sh) on CT 113 emits walg_last_success_timestamp_seconds + walg_archive_status to node-exporter textfile. Prometheus alert rules deployed on CT 109: WalgArchiveStale (>24h) + WalgArchiveFailed.
• Zoraxy audit — done 2026-05-23. All 22 routes verified; arcane updated to CT 109 backend; SkipWebSocketOriginCheck enabled on 13 routes; decommissioned routes noted. Missing: dozzle.nuclide.systems route (never created), immich-tools, paperless, paperless-ai. See services/zoraxy.md.

🟡 Medium — incomplete migrations / cleanup debt

• Docker disk reclaim — ~17 GB reclaimable on CT 104 (10.3 GB unused images, 6.9 GB build cache). docker system prune -a after confirming no needed images. See infra/proxmox-state.md §10c.
• Dormant stacks audit — 4 decommissioned 2026-05-23 (daytona, homepage, proxy, streamio — composes renamed .DECOMMISSIONED). Remaining ~6 still to audit: dozzle (CT 104 copy), arr-stack, qdrant, etc. See infra/proxmox-state.md §10c.
• Config-to-git — deployed 2026-05-23. Scripts + daily 03:00 crons on CT 108, CT 102, PVE. Pushes to fkrebs/zoraxy-conf, fkrebs/adguard-conf, fkrebs/pve-conf (force-push). HA (fkrebs/ha-config) still pending — needs Git Pull addon on HAOS.
• D-Link hardening — admin UI open to full LAN, SNMP unverified, HTTP-only. Enable trusted-host allowlist, harden SNMP, add TLS. See services/homelab-architecture.md.
• Vaultwarden OIDC SSO — Pocket-ID client created 2026-05-21, auth flow not wired up. Confirmed no OIDC vars in /opt/stacks/vaultwarden/.env as of 2026-05-22.

🟡 Medium — incomplete migrations / cleanup debt

• Infisical OIDC config — Pocket-ID client created (b2069075-ede2-4251-ad1f-9a62e6a188b3). Manual step needed: enter OIDC settings in Infisical admin at http://192.168.1.7:8200/admin → Settings → OIDC. Issuer: https://id.nuclide.systems, ClientID: b2069075, Secret: qsANw95zsza0_tzJR9SzSp4Z6iEIhidgJPHpDih1ffE.
• Infisical secrets migration — Phase 2: import CT104 .env keys. Use infisical import. See services/secrets-manager.md Phase 2.
• Arcane auto-update notifications — ntfy.nuclide.systems provider configured; auto-update daily 3am set. Gotify Apprise incompatible (payload format). Workaround: configure via Arcane UI or add shoutrrr plugin. Gotify app token: AQnDNB2Y6JlOewc (app: arcane-updates).
• Loki dashboard broken in Grafana — fixed 2026-05-23. K8s labels (namespace/pod) replaced with Docker labels (host/container/job) in 14 places across template variables and panel LogQL expressions. Provisioning file updated + pushed via Grafana API.
• Grafana HAOS dashboard — HA recorder reconfiguration planned (exclude Prometheus-covered metrics). Reduce recorder include list.
• HAOS SSH key — manual step: paste nuc public key in HA terminal.

🔵 Planned — requires infrastructure or significant effort

• Second NVMe — needed before rpool mirror + Postgres consolidation (CT 113). Blocks several items.
• Infisical secrets migration — CT 112 deployed (192.168.1.7:8200, LAN-only). Phase 1 done. Remaining:
  • Phase 2: create project homelab/ct104, import ai/.env (~55 keys) via infisical import, add agent sidecars to compose stacks
  • Phase 3: replace hardcoded env vars in Coder main.tf (committed to Gitea — security risk)
  • Phase 4: OIDC SSO via Pocket-ID (low priority — convenience) See services/secrets-manager.md.
• Consolidate syncstack → MCP gateway — done 2026-05-23. litellm_sync.py ported from syncstack, mounted into gateway. Gateway runs run_sync() every 15 min (120s startup delay). /etc/cron.d/syncstack deleted; Dockerfile.syncstack → .DECOMMISSIONED-2026-05-23; /opt/stacks/CLAUDE.md updated.
• ZFS ARC tuning — raise zfs_arc_max from 6.2 GiB → 16 GiB once CT 101 memory right-sized. See infra/proxmox-state.md §3.
• Document ingestion n8n workflow — Docling MCP deployed; n8n orchestration not built. Nextcloud/Paperless → Docling → embeddings → LobeChat KB. See services/doc-ingestion.md.
• ComfyUI async queue — MCP tool blocks 90–300 s. Job-queue pattern: generate_image() returns ID; get_job_status() polls. See services/comfyui.md.
• Additional MCP servers — Gitea ✓, Paperless ✓, Proxmox-VE ✓ (done 2026-05-23). Remaining: Karakeep (needs API key from hoarder UI), Vaultwarden, Audiobookshelf.
• Arcane + Dozzle → CT 109 — done 2026-05-23. Headless agents on all 7 Docker hosts with unique tokens per environment. 8 environments total: NUC, db, shepard, docker, nextcloud, id, dev, secrets. Auto-update daily 3am. ntfy notifications configured.
• crawl4ai SSE StreamConsumed — resolved 2026-05-22. Errors were from old code in log history; clean restart confirms no errors. Health: ok (7 tools), lobe-sync: 27/27 servers.

Recently completed (2026-05-23, session continued ×4)

• Docs consistency pass: zoraxy.md, arcane.md, ct-inventory.md, portmap.md, proxmox-memory-audit.md, mcp-gateway.md, RESUME.md all updated to reflect 2026-05-23 state.
• Wetty deployed on CT 109 (:4090, LAN-only). jump-menu.sh on nuc → SSH jump menu to all 10 hosts.
• MCP servers added: gitea, paperless, proxmox (read-only PVEAuditor role). Total: 30 servers.
• Loki dashboard fixed: K8s labels replaced with Docker labels in 14 places.
• WAL-G monitoring: textfile collector on CT 113 + Prometheus alert rules on CT 109.
• CT 103 RAM: 2 GB → 4 GB + 1 GB swap (Backrest OOM fix for 822 GB initial backup).
• Zoraxy WS: SkipWebSocketOriginCheck enabled on 13 routes. Arcane upstream corrected to CT 109.
• CT 104 cleanup: daytona, homepage, proxy, streamio decommissioned.
• Paperless-ngx: USERMAP_UID=997 + file permissions fixed, container stable.

Recently completed (2026-05-23, session continued ×3)

• Homarr board deployed: 6 category sections (AI, Monitoring, Files, Dev, Home, Identity & Admin), 27 apps with ping-based status, Mainsail included in Dev. Fixed schema: x_offset/y_offset on section rows + paired empty sections (y=0 initial, y=1/2 per category). Board nuclide set as home board.
• Arcane environments fixed: deleted stale shared-token environments; created 6 unique edge environments (shepard, docker, nextcloud, id, dev, secrets) via API with useApiKey=true. Each agent updated with unique token. All 8 environments online.
• Arcane API key seeded: ADMIN_STATIC_API_KEY from compose was never seeded (DB pre-existed). Inserted via argon2id hash. Key arc_d3357a65 now functional for automation.
• Infisical SITE_URL fixed: changed from https://secrets.nuclide.systems to http://192.168.1.7:8200 (LAN-only, no Zoraxy route).
• Infisical Pocket-ID client created: client b2069075-ede2-4251-ad1f-9a62e6a188b3, callback http://192.168.1.7:8200/api/v1/sso/oidc/callback. Manual OIDC config entry still needed in Infisical admin UI.

Recently completed (2026-05-23, session continued ×2)

• PVE LXC startup order fixed: CT103/104/105/112/113 had no startup order. Set: CT113=4, CT112=6, CT104=7, CT103=8, CT105=9. Full deterministic boot sequence now in place. Also enabled CT109 onboot=1 (was missing).
• KSM enabled on PVE host: kernel samepage merging active + persisted via ksm-enable.service. Zero-cost memory deduplication across LXC containers.
• 5-minute critical watchdog deployed: homelab-health-quick.timer on nuc; QUICK=1 mode checks LXC status + critical containers every 5 min. Auto-starts stopped LXCs, notifies Gotify for HAOS KVM issues.
• PVE backlog issues filed: fkrebs/docs #2 (NFS v3→v4), #3 (ZFS ARC 16 GiB), #4 (hugepages for KVM).

Recently completed (2026-05-23, session continued)

• HAOS KVM OOM root-caused and fixed: karakeep_chrome on CT 104 triggered host OOM → killed HAOS KVM twice. Applied: shm_size: '512m' to chrome, oom_score_adj=-300 on KVM PID, systemd timer protect-haos-kvm.timer (every 5 min) for persistence.
• Arcane + Dozzle migrated to CT 109: servers on CT 109 (:10002, :10001). Headless agents deployed on CT 101, 104, 105, 110, 111, 112, 113. Zoraxy route updated. docs-server also on CT 109 (:13080).
• Homarr + Grafana OIDC wired: clients created via Pocket-ID API (no manual UI steps). Homarr client 63a94e30, Grafana client 92d987d5 (PKCE). Both SSO-active.
• n8n flows tracked in Gitea: repo fkrebs/n8n-flows with daily sync cron (CT 104, 03:30). 10 automation ideas filed as Gitea issues. Interim health check bash/systemd on nuc (issue #1 = migrate to n8n).
• CT 104 ai repo cleaned: lobehub/data/ untracked from git, .gitignore fixed, litellm-config + mcp-gateway changes committed.
• zoraxy-conf repo: arcane proxy update committed and pushed (CT 108).

Recently completed (2026-05-23)

• Config-to-git crons: scripts + daily 03:00 crons deployed on CT 108 (Zoraxy → zoraxy-conf), CT 102 (AdGuard → adguard-conf), PVE (rsync /etc/pve → pve-conf). Force-push. All three verified with initial push.
• Syncstack → MCP gateway: litellm_sync.py (ported from syncstack.py) mounted into gateway container. _litellm_sync_loop() runs every 15 min in gateway's asyncio loop. Cron /etc/cron.d/syncstack deleted. Dockerfile.syncstack decommissioned. /opt/stacks/CLAUDE.md updated.

Recently completed (session continued ×2, 2026-05-22)

• crawl4ai MCP fixed: was returning 404 on /mcp (streamable-HTTP). Current 0.8.6 image already has MCP over SSE at /mcp/sse. Added SSE transport support to mcp-gateway/server.py: per-request SSE round-trip with auto-initialize handshake; transport: sse in hardcoded SERVERS + config.json. Gateway health now shows crawl4ai ok; lobe-sync shows 27/27 servers.
• syncstack cron fixed: missing cd /opt/stacks/ai && prefix caused "no configuration file" every 15 min since ~May 18. Fixed. Manual run confirmed 60 models including 5 Claude models; LobeChat recreated.

Recently completed (session continued, 2026-05-22)

• claude-max-bridge /v1/responses endpoint: implemented OpenAI Responses API with previous_response_id chaining. Root cause of "session already in use": Claude CLI leaves session JSONL files in an un-resumable "dequeued" state after each --print run. Fix: maintain conversation history server-side as formatted text; inject into system prompt for continuations; always use --no-session-persistence. Also fixed: assistant turns in --input-format stream-json require content as array-of-blocks not string (JS error otherwise). Assistant content array format also fixed in _parse_messages / _parse_responses_input.
• LiteLLM pass-through for /v1/responses: added pass_through_endpoints entry in /opt/stacks/ai/litellm-config/config.yaml → http://claude-max-bridge:8000/v1/responses. LiteLLM appends its token hash to the response id, so previous_response_id chaining only works when calling the bridge directly (not through LiteLLM).

Recently completed (this session, 2026-05-22)

• Pocket-ID backup: pre-hook on CT 103 runs pocket-id export + scp keys → UNAS staging; services-backup-plan snapshots → jottacloud. Verified.
• Stale NFS vaultwarden dir deleted: /mnt/pve/unas/services/vaultwarden/ removed.
• Nextcloud CIFS → NFS: crash-loop fixed; CT 105 mp0 migrated from /mnt/pve/unas_smb to /mnt/pve/unas; CIFS fstab entry removed.
• Vaultwarden → local zfs: /mnt/pve/unas/services/vaultwarden → /opt/stacks/vaultwarden/data; stale db.sqlite3 deleted; DB already on CT 113 postgres.
• WAL-G fixed: immich_postgres and lobe-postgres backups now succeed. Fixed: missing -u postgres, missing data path, set -e abort, dead shared-postgres reference.
• ComfyUI --async-offload removed: caused CPU fallback + pure-noise output on GGUF img2img with XPU.
• apps/ decommissioned: dead duplicate compose files in /opt/stacks/apps/ renamed .DECOMMISSIONED-2026-05-22.
• Docs consolidated: storage.md, volumes.md, homelab-architecture.md, proxmox-state.md, proxmox-memory-audit.md, ct-inventory.md, README.md all reconciled.

Key system state

Host	IP	Role
nuc (PVE)	192.168.1.20	Proxmox host — SSH gateway to all CTs; jump-menu.sh
CT 109 ops	192.168.1.8	Monitoring — Prometheus + Grafana + Loki + Arcane + Dozzle + Homarr + Wetty
CT 104 docker	192.168.1.40	Main Docker host — AI/ML, media, ~65 containers
CT 113 db	192.168.1.6	Shared Postgres 17 + WAL-G → Garage S3 (cron 02:00)
CT 110 id	192.168.1.5	Pocket-ID OIDC IdP
CT 103 backrest	192.168.1.3	Backrest — jottacloud via rclone; UNAS at /mnt/pve/unas
CT 111 dev	192.168.1.42	Coder + Gitea
CT 108 zoraxy	192.168.1.4	Reverse proxy + ACME — always confirm before changes

Constraints to remember

• Zoraxy (CT 108): always get explicit user confirmation before any config change.
• Infisical (CT 112): LAN-only, no Zoraxy route — must not be internet-exposed.
• CT 104 has a separate agent reconciling compose files — rename decommissioned files (don't just stop), update /opt/stacks/CLAUDE.md.
• ComfyUI container limit: 24 G. Do not raise without measuring host impact.
• Pocket-ID is SQLite-only — no Postgres migration possible.