Disaster Recovery

Overview

Rousseau's entire operational state fits in two SQLite files:

• sessions.db — conversations, cron jobs, JID map, FTS5 index, claude cache.
• whatsapp.db — whatsmeow device credentials (WhatsApp only).

Every DR strategy here revolves around backing those files up and restoring them cleanly.

Backup

D=$(date -u +%Y%m%d-%H%M%SZ)
mkdir -p ~/backup/rousseau/$D
sqlite3 ~/.local/share/rousseau/sessions.db ".backup ~/backup/rousseau/$D/sessions.db"
cp ~/.local/share/rousseau/whatsapp.db ~/backup/rousseau/$D/whatsapp.db 2>/dev/null || true

# ~/.config/systemd/user/rousseau-backup.service
[Service]
Type=oneshot
ExecStart=/usr/local/bin/rousseau-backup.sh

# ~/.config/systemd/user/rousseau-backup.timer
[Timer]
OnCalendar=daily
Persistent=true
[Install]
WantedBy=timers.target

# /usr/local/bin/rousseau-backup.sh
#!/usr/bin/env bash
set -euo pipefail
D=$(date -u +%Y%m%d-%H%M%SZ)
DEST="$HOME/backup/rousseau/$D"
mkdir -p "$DEST"
sqlite3 "$HOME/.local/share/rousseau/sessions.db" ".backup $DEST/sessions.db"
cp "$HOME/.local/share/rousseau/whatsapp.db" "$DEST/" 2>/dev/null || true
find "$HOME/backup/rousseau" -mindepth 1 -maxdepth 1 -mtime +30 -print0 | xargs -0 rm -rf --

export RESTIC_REPOSITORY=b2:rousseau-bucket:backups
export RESTIC_PASSWORD_FILE=/etc/rousseau/restic.pw

restic backup \
  ~/.local/share/rousseau/sessions.db \
  ~/.local/share/rousseau/whatsapp.db \
  --tag rousseau

Restore

# Stop any running rousseau processes first
systemctl --user stop 'rousseau-*.service'
pkill -TERM rousseau 2>/dev/null || true

# Restore
cp ~/backup/rousseau/<D>/sessions.db ~/.local/share/rousseau/
cp ~/backup/rousseau/<D>/whatsapp.db ~/.local/share/rousseau/ 2>/dev/null || true

# Verify
sqlite3 ~/.local/share/rousseau/sessions.db "PRAGMA integrity_check;"
# expect: ok

# Start
systemctl --user start 'rousseau-*.service'

Integrity checks

Run monthly:

sqlite3 ~/.local/share/rousseau/sessions.db "PRAGMA integrity_check;"
sqlite3 ~/.local/share/rousseau/sessions.db "PRAGMA quick_check;"
sqlite3 ~/.local/share/rousseau/sessions.db "SELECT COUNT(*) FROM sessions;"

If integrity_check returns anything other than ok, restore from the latest good backup.

What survives

Failure modes

• Backup taken during a heavy write — WAL means readers see a consistent snapshot, but a plain cp might catch the WAL mid-checkpoint. Use .backup or stop the daemon.
• Restore into a running daemon — silent corruption. Always stop the process first.
• Restore with mismatched user — the file will be owned by the wrong user after container/host role changes. chown back to rousseau:rousseau (or your uid).

Related pages

• Reference: Session store
• Reference: Config: State
• Migrations: Overview
• Best Practices: Session hygiene