Session 52: BUG-072 production outage fixed (ufw+Docker conflict + dual deployment)

projects/business/memory/bugs.md

@@ -1,3 +1,21 @@
+## BUG-072: Production Outage — UFW+Docker Conflict + Dual Deployment
+- **Date:** 2026-02-18 ~08:00 UTC
+- **Severity:** CRITICAL
+- **Duration:** Unknown start time → fixed 08:04 UTC
+- **Symptoms:** App container stuck in restart loop, unable to reach PostgreSQL. `ENETUNREACH 172.17.0.1:5432`
+- **Root causes:**
+  1. UFW injected DROP rules into Docker's DOCKER iptables chain, blocking container-to-host networking
+  2. A systemd service (`docfast.service`) running a separate Node.js process from `/opt/docfast` was binding port 3100, preventing Docker container from starting after fix
+- **Fix applied:**
+  - Removed iptables DROP rules from DOCKER chain
+  - Stopped and disabled `docfast.service` systemd unit
+  - Changed DATABASE_HOST from `172.17.0.1` to `host.docker.internal` in docker-compose.yml
+  - Full `docker compose down && up -d` — container healthy
+- **Permanent fix:** DevOps agent dispatched to properly resolve ufw+Docker conflict and remove dual deployment
+- **Status:** FIXED (immediate), permanent fix in progress
+
+---
+
 # DocFast QA Report — 2026-02-15
 
 **Tester:** QA Bot (automated)