155 lines
15 KiB
HTML
155 lines
15 KiB
HTML
<!DOCTYPE html>
|
|
<html lang="en">
|
|
<head>
|
|
<meta charset="UTF-8">
|
|
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
|
<title>Frank Program — Weekend Report & Doc Audit (2026-08-23)</title>
|
|
<style>
|
|
:root{
|
|
--bg:#0d1117; --panel:#161b22; --panel2:#1c2330; --border:#30363d;
|
|
--fg:#e6edf3; --dim:#8b949e; --green:#3fb950; --red:#f85149;
|
|
--yellow:#d29922; --blue:#58a6ff; --purple:#bc8cff; --mono:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace;
|
|
}
|
|
*{box-sizing:border-box}
|
|
body{margin:0;background:var(--bg);color:var(--fg);font:16px/1.6 -apple-system,BlinkMacSystemFont,"Segoe UI",Helvetica,Arial,sans-serif;}
|
|
.wrap{max-width:1040px;margin:0 auto;padding:32px 20px 80px;}
|
|
header.hero{background:linear-gradient(135deg,#161b22,#1c2330);border:1px solid var(--border);border-radius:12px;padding:28px 30px;margin-bottom:28px;}
|
|
header.hero h1{margin:0 0 6px;font-size:26px;letter-spacing:.2px;}
|
|
header.hero .sub{color:var(--dim);font-size:15px;}
|
|
header.hero .meta{margin-top:14px;display:flex;gap:10px;flex-wrap:wrap;}
|
|
.pill{background:var(--panel2);border:1px solid var(--border);border-radius:999px;padding:4px 12px;font-size:13px;color:var(--dim);}
|
|
.pill b{color:var(--fg);}
|
|
h2{font-size:20px;margin:38px 0 14px;padding-bottom:8px;border-bottom:1px solid var(--border);color:var(--blue);}
|
|
h3{font-size:16px;margin:22px 0 8px;color:var(--purple);}
|
|
p{margin:10px 0;}
|
|
code{font-family:var(--mono);background:var(--panel2);border:1px solid var(--border);border-radius:4px;padding:1px 6px;font-size:13.5px;color:#c9d1d9;}
|
|
ul,ol{margin:8px 0;padding-left:24px;} li{margin:4px 0;}
|
|
table{width:100%;border-collapse:collapse;margin:14px 0;font-size:14px;}
|
|
th,td{border:1px solid var(--border);padding:8px 10px;text-align:left;vertical-align:top;}
|
|
th{background:var(--panel2);color:var(--fg);font-weight:600;}
|
|
tr:nth-child(even) td{background:rgba(255,255,255,.015);}
|
|
.tag{display:inline-block;border-radius:4px;padding:1px 8px;font-size:12px;font-weight:600;font-family:var(--mono);}
|
|
.ok{background:rgba(63,185,80,.15);color:var(--green);border:1px solid rgba(63,185,80,.4);}
|
|
.warn{background:rgba(210,153,34,.15);color:var(--yellow);border:1px solid rgba(210,153,34,.4);}
|
|
.err{background:rgba(248,81,73,.15);color:var(--red);border:1px solid rgba(248,81,73,.4);}
|
|
.info{background:rgba(88,166,255,.15);color:var(--blue);border:1px solid rgba(88,166,255,.4);}
|
|
.card{background:var(--panel);border:1px solid var(--border);border-radius:10px;padding:18px 20px;margin:14px 0;}
|
|
.grid{display:grid;grid-template-columns:repeat(auto-fit,minmax(180px,1fr));gap:12px;margin:16px 0;}
|
|
.stat{background:var(--panel);border:1px solid var(--border);border-radius:10px;padding:16px;text-align:center;}
|
|
.stat .n{font-size:30px;font-weight:700;color:var(--green);font-family:var(--mono);}
|
|
.stat .l{color:var(--dim);font-size:13px;margin-top:4px;}
|
|
.note{border-left:3px solid var(--blue);background:var(--panel);padding:10px 16px;border-radius:0 8px 8px 0;margin:14px 0;color:var(--dim);}
|
|
.done li::marker{color:var(--green);}
|
|
.todo li::marker{color:var(--yellow);}
|
|
footer{margin-top:48px;color:var(--dim);font-size:13px;border-top:1px solid var(--border);padding-top:16px;}
|
|
a{color:var(--blue);text-decoration:none;} a:hover{text-decoration:underline;}
|
|
.small{font-size:13px;color:var(--dim);}
|
|
</style>
|
|
</head>
|
|
<body>
|
|
<div class="wrap">
|
|
|
|
<header class="hero">
|
|
<h1>Frank Program — Weekend Report & Documentation Audit</h1>
|
|
<div class="sub">serverAppConfigs · the two-host homelab (Dankserver 192.168.0.235 · Dankclaw 192.168.0.143)</div>
|
|
<div class="meta">
|
|
<span class="pill">Date <b>2026-08-23 (Sun)</b></span>
|
|
<span class="pill">Window <b>2026-08-22 → 08-23</b></span>
|
|
<span class="pill">Tests <b>116 / 116 pass</b></span>
|
|
<span class="pill">Broken links <b>34 → 0</b></span>
|
|
<span class="pill">Auditors <b>5 subagents</b></span>
|
|
</div>
|
|
</header>
|
|
|
|
<h2>Executive summary</h2>
|
|
<p>This weekend consolidated the LLM-fleet repo into <code>serverAppConfigs</code>, moved both LLM
|
|
services to port <code>7770</code>, built and deployed the <b>Homepage</b> (TDD, terminal-TUI aesthetic),
|
|
reset the homepage login, and then ran a <b>five-subagent documentation audit</b> that fact-checked the
|
|
knowledge base against the live system, researched each service's official setup docs online, and verified
|
|
progress. The audit's findings are distilled into the KB now; the remaining work is listed at the end.</p>
|
|
|
|
<div class="grid">
|
|
<div class="stat"><div class="n">34 → 0</div><div class="l">broken links fixed</div></div>
|
|
<div class="stat"><div class="n">0</div><div class="l">.AGENTTRASH links (none existed)</div></div>
|
|
<div class="stat"><div class="n">5</div><div class="l">auditor subagents fanned out</div></div>
|
|
<div class="stat"><div class="n">~20</div><div class="l">stale facts corrected in the KB</div></div>
|
|
<div class="stat"><div class="n">116</div><div class="l">tests pass (all seams)</div></div>
|
|
<div class="stat"><div class="n">2</div><div class="l">LLM services healthy on :7770</div></div>
|
|
</div>
|
|
|
|
<h2>Weekend progress</h2>
|
|
<ul class="done">
|
|
<li><b>Repo merge.</b> <code>vllmScripts</code> merged into <code>serverAppConfigs</code> with full history preserved (two-parent import under the retired <code>llm/</code> prefix, then reorganized). One repo, one checkout per host, one deploy script.</li>
|
|
<li><b>Port migration.</b> <code>frank-default-large</code> moved <code>52625 → 7770</code> (Dankclaw). Both hosts now serve on <code>:7770</code> (different IPs, no conflict). Client (<code>models.json</code>) updated first, then the server, for a clean restart.</li>
|
|
<li><b>Homepage (grilled → built → deployed).</b> 43 design questions settled. TDD across 6 slices (checker core, HTTP probe, real <code>services.json</code> + schema, pages/theme/toggle/footer, Caddy auth split, git-driven deploy). Live on <code>https://dankclaw.franksdank.cloud/</code>; <code>/admin/</code> is operator-only.</li>
|
|
<li><b>Homepage login reset.</b> User page + shared family password now <code>dankuser</code> / <code>dankpassword.123</code> (a dedicated <code>CADDY_HOMEPAGE_HASH</code>); the operator <code>/admin/</code> page still uses <code>frank</code>.</li>
|
|
<li><b>Caddy KB.</b> Added <code>~/.pi/agent/docs/caddy/</code> (Caddyfile reference + index), cross-linked from the root index.</li>
|
|
<li><b>Documentation audit (this session).</b> Fanned out 5 read-only <code>doc-auditor</code> subagents; fixed all 34 broken links; corrected ~20 stale facts; verified every service against the live system. Committed <code>90c4ac6</code>.</li>
|
|
</ul>
|
|
|
|
<h2>Doc audit — method</h2>
|
|
<p>A custom <code>docaudit.doc-auditor</code> subagent (read-only on the repo; <code>bash</code> for read-only
|
|
<code>ssh</code>, <code>web_search</code> for the official docs, <code>write</code> for a scratch findings file) ran
|
|
five times — one per service group. Each audited its pages three ways: <b>1) fact-check</b> the concrete claims
|
|
against the real config + live system, <b>2) online research</b> the official setup/config docs, <b>3) progress
|
|
verify</b> (deployed vs TODO vs retired). Findings were written to scratch files; I (the parent) synthesized and
|
|
applied them as the sole writer. The link inventory was produced deterministically up front (a Python resolver for
|
|
wikilinks + relative markdown links).</p>
|
|
|
|
<div class="note">The <code>.AGENTTRASH</code> dirs hold only subagent artifacts (no doc pages), and <b>no doc links to any trashed page</b>. The four <code>.AGENTTRASH</code> references in the docs are accurate descriptive text (paths in backticks), not links — so nothing to sever there.</div>
|
|
|
|
<h2>Findings by group</h2>
|
|
|
|
<h3>1 · LLM fleet — active production <span class="tag ok">verified</span></h3>
|
|
<table>
|
|
<tr><th>Service</th><th>State</th><th>Verified against</th></tr>
|
|
<tr><td><code>frank-default-small</code> (Dankserver :7770)</td><td><span class="tag ok">ACTIVE</span></td><td>llama.cpp <code>b10121</code>, HauhauCS Qwen3.6-35B-A3B Q6_K_P + ngram-mod, <code>{"status":"ok"}</code></td></tr>
|
|
<tr><td><code>frank-default-large</code> (Dankclaw :7770)</td><td><span class="tag ok">ACTIVE</span></td><td>CIRU ROCmFPX v2.3+patch, jcbtc Qwen3.8-27B CIRU v3, 262144 ctx / 1 slot, <code>{"status":"ok"}</code></td></tr>
|
|
</table>
|
|
<p class="small">Corrected: LLM index <code>36 → 38</code> MoE experts + <code>700K → 786432</code> ctx; the small-benchmark "Chosen State" now flags the live drift (<code>ncmoe 38 / batch 4096 / ubatch 2048</code>); <code>frank-default-small</code> marked active (not <code>frank-pp-worker</code>); era-notes added to the large-benchmark + spec-matrix pages (PP-era, superseded by CIRU v3); the two vLLM-focused architecture pages now carry a "current runtime is llama.cpp" header.</p>
|
|
|
|
<h3>2 · LLM fleet — experiments / research <span class="tag info">retired history</span></h3>
|
|
<p class="small">Verified the retired/superseded status of the whole experimental corpus: Pipeline Parallelism (retired 2026-08-17), DFlash2-Vulkan (archived 08-21), Kairic Edge (reverted 08-22), vLLM (retired 08-18), FastMTP (retired 08-18), FastFlowLM NPU (retired 08-17). Added a <b>RETIRED</b> banner to the vLLM index; corrected the DFlash page (frank-default-small is active, not retired). Online research: vLLM spec-decoding (PP V2), DFlash/DSpark, Kairic IU4, GGUF quantization trade-offs, llama.cpp 262K context.</p>
|
|
|
|
<h3>3 · Media stack (Dankserver) <span class="tag warn">attention</span></h3>
|
|
<table>
|
|
<tr><th>Page</th><th>Correction</th><th>Severity</th></tr>
|
|
<tr><td><code>readarr.md</code></td><td><b>Readarr is RETIRED</b> (May 2026, Servarr Team — unusable metadata, stalled Open Library migration). Added a banner; alternatives: Bookshelf fork, Calibre content server.</td><td><span class="tag err">HIGH</span></td></tr>
|
|
<tr><td><code>radarr.md</code></td><td>v4 → <b>v6</b> stable (v6.3.x).</td><td><span class="tag warn">MED</span></td></tr>
|
|
<tr><td><code>Calibre.md</code></td><td>9.9.0 → <b>9.2.1</b> (live host).</td><td><span class="tag warn">MED</span></td></tr>
|
|
<tr><td><code>Calibre-Web.md</code></td><td>0.6.27-ls397 → <b>0.6.25-ls365</b> (live).</td><td><span class="tag warn">MED</span></td></tr>
|
|
<tr><td><code>Copyparty.md</code></td><td>v1.20.19 → <b>v1.20.21</b>.</td><td><span class="tag warn">MED</span></td></tr>
|
|
<tr><td><code>Dankserver Stack.md</code></td><td>Caddy is on <b>Dankclaw</b> (not "the host" = Dankserver).</td><td><span class="tag warn">MED</span></td></tr>
|
|
</table>
|
|
<div class="note"><b>Open discrepancy.</b> The live check shows only <code>calibre-web</code> running as a podman container; RomM, Copyparty, Jellyfin, the *arr stack, MariaDB, and Redis are <b>not deployed</b>. The operator previously recorded the media stack as "UP (bare processes) — not an outage, separate thread." Re-verify that state before treating the other services as live. See "Remaining work".</div>
|
|
|
|
<h3>4 · AI / search / dev stack (Dankclaw) <span class="tag ok">verified</span></h3>
|
|
<p class="small">All services <span class="tag ok">DEPLOYED</span> + healthy (SearXNG trio, ComfyUI GPU, OpenWebUI, Hermes + dashboard, Gitea trio, Postgres, Valkey, Gluetun). Corrected: the <b>misleading hostnames</b> (docs said <code>*.dankclaw.local</code>; Caddy actually serves <code>*.franksdank.cloud</code>) on SearXNG + the stack page; the Hermes gotcha host (<code>frank-default-large</code> is on Dankclaw, not Dankserver); the Service Research Brief (ComfyUI CPU→<b>GPU</b> via gfx1151; OpenWebUI endpoint <code>host.containers.internal:7770</code> → <code>192.168.0.143:7770</code>); the boot-persistence table (ComfyUI + OpenWebUI are boot-persistent; Hermes is compose-only). Online research: SearXNG, ComfyUI ROCm, OpenWebUI, Hermes, Gitea, Gluetun, Valkey, Postgres.</p>
|
|
|
|
<h3>5 · Infrastructure / observability / architecture <span class="tag ok">verified</span></h3>
|
|
<p class="small">Caddy, Homepage, Grafana, Prometheus, Loki, Alloy, Node-Exporter, LAN-DNS all <span class="tag ok">DEPLOYED</span>. Corrected: Grafana <code>GF_SERVER_ROOT_URL</code> + hostname (<code>https://grafana.franksdank.cloud</code>); the Loki + Alloy images (<code>franksdank/…</code> local builds with a busybox healthcheck); the Observability mermaid diagram (<code>Promtail → Alloy</code>); the Homepage Grilling test count (112 → 116). Online research: Caddy (<code>handle</code> ordering, <code>hash-password</code> variants), Prometheus (9 live scrape jobs), Loki (v13 TSDB, 720h retention), Alloy (River, replaces Promtail EOL 2026-03-02), Node-Exporter, dnsmasq quadlet.</p>
|
|
|
|
<h2>What else needs to be done</h2>
|
|
<ol class="todo">
|
|
<li><b>Re-verify the media-stack state.</b> The live check shows only <code>calibre-web</code> up; the other services are not deployed as containers. Confirm whether they run as bare processes or are down, then reconcile <code>Dankserver Stack.md</code>. <span class="tag warn">separate thread</span></li>
|
|
<li><b>Replace Readarr.</b> It is retired upstream. Decide between the Bookshelf community fork and the Calibre content server; update the *arr docs + compose.</li>
|
|
<li><b>Deploy the *arr stack.</b> The docs describe Radarr/Sonarr/Lidarr/Readarr/Prowlarr/qBittorrent, but none are running. Bring them up (or mark them planned).</li>
|
|
<li><b>Per-service <code>/docs/<svc>/</code> pages.</b> User-facing doc pages on the Homepage (settled in the grilling; built incrementally).</li>
|
|
<li><b>Finish the online-research distill.</b> The auditors gathered the official setup notes; the fact-check fixes are applied. Add concise "Upstream setup" sections to the key service pages (RomM, Calibre-Web, SearXNG, ComfyUI, OpenWebUI, Hermes, Gitea, Loki/Alloy).</li>
|
|
<li><b>Hermes boot persistence.</b> <code>hermes</code> + <code>hermes-dashboard</code> are compose-only (<code>restart: unless-stopped</code>, no systemd unit). Add a boot-persistent unit to match the rest of the stack.</li>
|
|
<li><b>Alloy health endpoint.</b> The UI listens on <code>:12345</code> inside the container (not published); the status checker may report it "down". Publish the port or point the probe at a real health path.</li>
|
|
<li><b>Optional <code>/how/</code> page</b> on the Homepage (settled as optional in the grilling).</li>
|
|
<li><b>Grafana v13 review.</b> Running v13.2.0 (the 2026 major release). Review the breaking changes + the new Git-Sync / Assistant features.</li>
|
|
</ol>
|
|
|
|
<h2>Audit artifacts</h2>
|
|
<p class="small">The five auditor findings files (full fact-check tables, online-research distills, and per-page status) are in the run's artifact dir: <code>.pi-subagents/artifacts/outputs/a337bf24-…/.docaudit/</code> (<code>llm-active</code>, <code>llm-research</code>, <code>media</code>, <code>ai</code>, <code>infra</code>). The <code>doc-auditor</code> agent is registered at <code>~/.agents/docaudit.doc-auditor.md</code> for reuse. The <code>.docaudit/</code> scratch dir is git-ignored.</p>
|
|
|
|
<footer>
|
|
Frank Program · main designer's report · generated 2026-08-23 · source of truth: <code>serverAppConfigs</code> @ <code>90c4ac6</code> · all 116 tests pass
|
|
</footer>
|
|
|
|
</div>
|
|
</body>
|
|
</html>
|