Runbook: Backup & Restore (DR)

Synced read-only from /home/synnet/synvirt-platform/docs-internal/runbooks/backup-restore.md. Edit at the source, not here.

Runbook: Backup & Restore (DR)

Status: draft for M8. The restore path MUST be tested before launch.

What to back up

Asset	Method	Destination
Postgres (per DB)	pg_dump per database, scheduled	off-server (offsite)
Host volumes	Veeam (filesystem level)	per existing ops
Discourse uploads	included in its volume/backup	offsite
Payload media	local uploads dir (web app volume) — filesystem backup	offsite
Keycloak realm	exported JSON in repo (infra/keycloak/)	git

Per-database dump (example)

for db in portal payload keycloak discourse; do
  docker compose -f infra/compose/compose.yaml exec -T postgres \
    pg_dump -U "$POSTGRES_USER" -Fc "$db" > "backup/${db}-$(date +%F).dump"
done
# ship backup/ offsite (rsync/rclone to external target)

Restore a single database

# recreate role/db if needed (infra/postgres/init runs only on first boot)
docker compose -f infra/compose/compose.yaml exec -T postgres \
  pg_restore -U "$POSTGRES_USER" -d portal --clean --if-exists < backup/portal-YYYY-MM-DD.dump

Migratability note

Because each service owns its database, a single restore moves exactly one service’s data — no shared-schema untangling. Same dumps can seed a dedicated data host during the single→multi-host split.

Pre-launch checklist

• Scheduled pg_dump job verified (files land offsite)
• Payload media (uploads dir) covered by the filesystem backup
• Full restore rehearsed into a scratch environment
• Recovery time + steps documented here