Backups, Restores, and Disaster Recovery

1. 1

   Logical backup and restore with pg_dump

   Start Postgres in Docker, create data, dump it in custom format (compressed, parallel-restorable), drop the table, and restore.

   terminalcopy cmd

   $ docker run -d --name pg -e POSTGRES_PASSWORD=pw -p 5432:5432 postgres:17

   export PGHOST=localhost PGUSER=postgres PGPASSWORD=pw

   psql -c "create table orders(id serial primary key, total numeric, created_at timestamptz default now())"

   psql -c "insert into orders(total) select random()*100 from generate_series(1,100000)"

   pg_dump -Fc -f shop.dump postgres && ls -lh shop.dump

   psql -c 'drop table orders'

   pg_restore -d postgres shop.dump && psql -c 'select count(*) from orders'

   ── expected output ──

   -rw-r--r-- 1 you you 2.1M shop.dump

   count

   --------

   100000

2. 2

   Point-in-time recovery on RDS

   RDS keeps automated backups plus transaction logs for the retention period (1–35 days), so you can restore to any second. A restore always creates a NEW instance; you then repoint the app or copy the missing rows back. Terraform course Stage 4 created this database with backup_retention_period set.

   terminalcopy cmd

   $ aws rds describe-db-instances --db-instance-identifier shoplite-prod --query 'DBInstances[0].LatestRestorableTime'

   aws rds restore-db-instance-to-point-in-time \

   --source-db-instance-identifier shoplite-prod \

   --target-db-instance-identifier shoplite-prod-pitr \

   --restore-time 2026-09-27T08:32:59Z

   ── expected output ──

   "2026-09-27T08:41:10+00:00"

   { "DBInstance": { "DBInstanceIdentifier": "shoplite-prod-pitr", "DBInstanceStatus": "creating", ... } }

3. 3

   Velero: back up a namespace, restore it elsewhere

   After installing Velero with an S3 bucket and snapshot provider, a scheduled backup runs daily. Restoring into a new namespace is a safe way to test.

   terminalcopy cmd

   $ velero schedule create shop-daily --schedule '0 2 * * *' --include-namespaces shoplite-prod --ttl 720h

   velero backup create shop-now --from-schedule shop-daily --wait

   velero restore create --from-backup shop-now --namespace-mappings shoplite-prod:shoplite-restore-test --wait

   kubectl -n shoplite-restore-test get pods,pvc

   ── expected output ──

   Backup completed with status: Completed.

   Restore completed with status: Completed.

   NAME READY STATUS

   pod/web-7c9d8e6f5-abcde 1/1 Running

   ...

   persistentvolumeclaim/data-redis-0 Bound

Each drill is a real failure mode. Read the scenario and the output, decide what's wrong, then reveal the diagnosis.