Add postmortem for duplicate email accounts issue

docs/postmortem-2026-01-29-duplicate-email-accounts.md

@@ -0,0 +1,85 @@
+# Postmortem: Duplicate User Accounts from Case-Sensitive Email Handling
+
+**Date**: 2026-01-29
+**Severity**: Medium
+**Status**: Resolved
+
+## Summary
+
+Users reported receiving duplicate email notifications. Investigation revealed multiple user accounts existed with the same email address differing only by case (e.g., `user@example.com` and `User@Example.com`).
+
+## Impact
+
+- 6 users affected with duplicate accounts
+- Duplicate email notifications sent to affected users
+- User data split across multiple accounts
+
+## Root Cause
+
+The email field in the User model used case-sensitive matching for both:
+1. Uniqueness constraint validation
+2. User lookup during authentication
+
+This allowed users to create multiple accounts by varying the case of their email address during signup.
+
+## Timeline
+
+- **2026-01-29 08:00** - User report of duplicate notifications received
+- **2026-01-29 08:15** - Investigation began via production database
+- **2026-01-29 08:30** - Root cause identified: case-sensitive email handling
+- **2026-01-29 08:45** - Fix developed and tested locally
+- **2026-01-29 09:00** - Fix deployed to production
+- **2026-01-29 09:15** - Merge script created to consolidate duplicate accounts
+- **2026-01-29 09:30** - All duplicate accounts merged, issue resolved
+
+## Resolution
+
+### Code Changes
+
+1. **Email normalization on creation** (`remarkbox/models/user.py`)
+   - Emails now stored as lowercase in `User.__init__`
+
+2. **Case-insensitive lookup** (`remarkbox/models/user.py`)
+   - `get_user_by_email()` now uses `func.lower()` for comparison
+
+3. **Merge script** (`remarkbox/scripts/merge_duplicate_email_users.py`)
+   - New management command to find and merge duplicate accounts
+   - Keeps oldest account, transfers all related data
+   - Handles unique constraint conflicts gracefully
+
+### Data Migration
+
+Ran `remarkbox_merge_duplicate_email_users` to consolidate:
+- 6 duplicate email sets merged
+- 14 nodes transferred
+- 4 watchers transferred
+- 1 notification transferred
+- 6 duplicate accounts deleted
+- All emails normalized to lowercase
+
+## Prevention
+
+1. **Unit tests added** (`test_models.py`)
+   - `test_email_normalized_to_lowercase`
+   - `test_email_mixed_case_normalized`
+
+2. **Integration tests added** (`test_views.py`)
+   - `test_get_user_by_email_case_insensitive`
+   - `test_get_or_create_returns_existing_regardless_of_case`
+   - `test_new_user_email_stored_lowercase`
+   - `test_no_duplicate_accounts_from_case_variations`
+
+## Lessons Learned
+
+1. Email addresses should always be normalized to lowercase on input
+2. Case-insensitive comparison should be used for email lookups
+3. Database unique constraints alone don't prevent case-variant duplicates in SQLite
+
+## Action Items
+
+- [x] Fix email normalization in User model
+- [x] Fix case-insensitive email lookup
+- [x] Create merge script for existing duplicates
+- [x] Add regression tests
+- [x] Run merge on production
+- [x] Document in postmortem