docfast: session 58 — 12 a11y/SEO fixes deployed, old server down (BUG-078)

projects/business/memory/bugs.md

@@ -1,3 +1,17 @@
+## BUG-078: Old Server Down — SMTP Relay + CI Runner Broken
+- **Date:** 2026-02-19 14:09 UTC
+- **Severity:** CRITICAL
+- **Issue:** Old server (167.235.156.214) is completely unreachable — 100% packet loss. This server runs:
+  1. **Postfix SMTP relay** with DKIM for docfast.dev — ALL signup verification and key recovery emails route through it
+  2. **Forgejo Actions CI runner** — CI jobs stuck in "pending", no new images built
+- **Impact:** 
+  - New signups CANNOT receive verification emails → no new customers
+  - Code changes cannot be built/deployed through CI pipeline
+  - Commit 37386bf stuck in pending CI
+- **Workaround:** Production manually updated to fb05989 image (accessibility fixes). But no email capability.
+- **Fix needed:** Investor must reboot old server via Hetzner Console. Long-term: migrate SMTP to K3s cluster and CI runner to K3s.
+- **Status:** OPEN — escalated to investor
+
 ## BUG-077: Cannot Push Code — Forgejo SSH Down + Token Lacks Write Scope
 - **Date:** 2026-02-19 13:15 UTC
 - **Severity:** HIGH

projects/business/memory/sessions.md

@@ -1404,3 +1404,41 @@
 **Action needed from investor:**
 1. **Option A (quickest):** Log into Forgejo (git.cloonar.com) → Settings → Applications → Create token with `write:repository` scope → Update `FORGEJO_TOKEN` in `.credentials/services.env`
 2. **Option B:** Register the deploy key (SSH public key from `/tmp/docfast_key.pub` on k3s-mgr) in the docfast repo's deploy keys with write access
+
+## Session 58 — 2026-02-19 14:09 UTC (Afternoon Session)
+- **ACCESSIBILITY + SEO FIXES DEPLOYED TO PRODUCTION:**
+  - Frontend dev agent completed commit fb05989 with 12 bug fixes
+  - Deployed to staging via CI, verified, then manually set production image (CI runner down)
+  - Fixes verified on production:
+    - BUG-062 ✅ `<main>` now wraps all content (hero through pricing)
+    - BUG-063 ✅ EU heading changed from h3 to h2
+    - BUG-064 ✅ All modal inputs have aria-label attributes
+    - BUG-065 ✅ Close buttons have aria-label="Close"
+    - BUG-066 ✅ Modals have aria-modal="true"
+    - BUG-067 ✅ Skip-to-content with id="main-content" on main
+    - BUG-068 ✅ Hash detection for #change-email in app.js
+    - BUG-057 ✅ JSON-LD Pro description matches page
+    - BUG-059 ✅ /docs page gets twitter:image meta
+    - BUG-055 ℹ️ No duplicates found (already clean)
+    - BUG-056 ℹ️ Sitemap already correct (sitemaps.org)
+    - BUG-058 ℹ️ twitter:image already present on homepage
+  - BUG-051/052 ALSO RESOLVED: Duplicate HTTP headers gone on production
+- **CRITICAL: OLD SERVER (167.235.156.214) DOWN (BUG-078):**
+  - Completely unreachable — SSH timeout
+  - SMTP relay broken → NO signup verification emails can be sent
+  - Forgejo Actions CI runner was on this server → CI jobs stuck in "pending"
+  - Commit 37386bf (from backend agent) in repo but CI can't build it
+  - **New signups are BLOCKED until server is back**
+- **BUG-077 STILL OPEN:** Forgejo SSH port 2222 also down
+- **Support:** Zero open tickets ✅
+- **Production health:** All pods running, 2 workers healthy, DB good, ~135ms response
+- **Investor Test:**
+  1. Trust with money? ⚠️ No — email verification broken, can't onboard new customers
+  2. Data loss on crash? ✅ No
+  3. Free tier abuse? ✅ Rate limited
+  4. Lost key recovery? ⚠️ No — recovery emails can't send
+  5. Features match website? ✅ Yes
+- **Budget:** €181.71 remaining, Revenue: €9
+- **Open bugs:** 1 CRITICAL (BUG-078 server down), 1 HIGH (BUG-077 git access)
+- **Status:** NOT LAUNCH-READY — email infrastructure down. Existing customers unaffected (API works), but no new signups possible.
+- **Escalation:** Investor must reboot old server via Hetzner Console. Long-term: migrate SMTP + CI to K3s cluster to eliminate single point of failure.

projects/business/memory/state.json

@@ -3,7 +3,7 @@
   "phaseLabel": "Build Production-Grade Product",
   "status": "launch-ready",
   "product": "DocFast \u2014 HTML/Markdown to PDF API",
-  "currentPriority": "Git push blocked (BUG-077). 4 code fixes prepared locally. Need Forgejo SSH or write token.",
+  "currentPriority": "CRITICAL: Old server down → SMTP broken (no emails) + CI runner dead. 12 accessibility/SEO bugs fixed on production. Need old server rebooted.",
   "ownerDirectives_PRIORITY": "Process these IN ORDER. Do not skip.",
   "ownerDirectives": [
     "Stripe: owner has existing Stripe account from another project \u2014 use same account, just create separate Product + webhook endpoint for DocFast.",
@@ -88,15 +88,19 @@
     "NEVER_READ_DIRECTLY": true
   },
   "openBugs": {
-    "CRITICAL": [],
+    "CRITICAL": [
+      "BUG-078: Old server (167.235.156.214) down — SMTP relay broken (no emails) + CI runner dead"
+    ],
     "HIGH": [
       "BUG-077: Forgejo SSH port 2222 down + API token lacks write:repository scope"
     ],
     "MEDIUM": [],
     "LOW": [],
-    "note": "Session 57: Git push blocked. Changes ready locally."
+    "note": "Session 58: 12 a11y/SEO bugs fixed on prod. Old server down breaks email + CI."
   },
-  "blockers": [],
+  "blockers": [
+    "Old server reboot needed (investor action) — SMTP + CI depend on it"
+  ],
   "startDate": "2026-02-14",
-  "sessionCount": 60
+  "sessionCount": 58
 }