Backup Essentials

NetBox Enterprise makes two kinds of backup:

A PostgreSQL backup alone does not protect you against the loss of the host, because it stays in the cluster. You need both. Your only task is to set up the disaster recovery backups.

1. Set up disaster recovery backups

Do these steps once.

1. In the admin console, open Disaster Recovery, and then click Backup settings.
2. Enter your S3 bucket, credentials, endpoint, and region. If you have more than one NetBox Enterprise cluster, give each one its own Prefix.
3. Click Update storage settings.
4. Click the Settings & Schedule tab, select Enable scheduled backups, pick Daily or Weekly, and then click Update schedule.
5. Click the Backups tab, and then click Start a backup. Wait for it to show Completed.
6. Write down one number from your NetBox, such as the count on the Sites page. You compare it after a disaster recovery restore.

The schedule sets how much work you can lose. A daily schedule can lose up to one day of changes.

Each disaster recovery backup is kept for one month by default. The Retention policy card on the Settings & Schedule tab changes that.

2. Check that both kinds of backup work

Do this check after the setup, and again after each upgrade.

1. In the admin console, open Disaster Recovery, and then click the Backups tab. The newest disaster recovery backup must show Completed.

2. On the host, confirm that a PostgreSQL backup ran:

   kubectl -n kotsadm exec -it \
     "$(kubectl -n kotsadm get pod \
          -l postgres-operator.crunchydata.com/pgbackrest-dedicated= \
          -o name | head -1)" \
     -c pgbackrest -- pgbackrest info --stanza=db

   The output must list at least one full backup with a recent timestamp. No stanzas exist in the repository. means that no PostgreSQL backup ran yet. The first one runs on the next schedule time, which is 01:00 by default.

If one of the checks fails, see Verify that both kinds of backup work.

3. Restore on a new host

Use these steps when you lost the host, or when you want to test a disaster recovery restore.

1. Start with a host that has no cluster. To reuse the same host, run sudo ./netbox-enterprise reset --yes first. The host reboots, so connect again when it is up.
2. Download the installer for the NetBox Enterprise version that ran when the disaster recovery backup was made. The restore prompt shows only the disaster recovery backups that match the installer.
3. Run sudo ./netbox-enterprise restore, and enter the S3 settings from step 1.
4. When the installer shows the name of a disaster recovery backup, confirm that it is the one you expect, and then enter Y.
5. When the installer asks, enter continue. After a few minutes it shows Application restored!
6. Wait a few more minutes. Log in to NetBox, open the Sites page, and compare the count with the number you wrote down.

If NetBox shows no data, or PostgreSQL does not start, the PostgreSQL restore did not complete. See Disaster recovery restore procedure.

For a disaster recovery restore that adds nodes, see High-availability restore.

Four facts you must know

1. Use the matching installer. The restore prompt hides each disaster recovery backup that does not match the installer version, and it shows no message. See Confirm the installer matches your newest disaster recovery backup.
2. A disaster recovery backup from before 2.3.0 has no database. After you upgrade from an earlier release, take a PostgreSQL backup on demand, and then take a disaster recovery backup. Do not use the older ones. See Upgrades.
3. A running cluster does not prove that your data came back. Only the number from step 6 does. See Verify.
4. Do not turn off PostgreSQL backups. Without them, a disaster recovery backup has no database. See With PostgreSQL backups off, nothing backs up the database.

See Also

• Backup and Restore in Depth - Every check and every edge case, and manual backups
• PostgreSQL Backups and Disaster Recovery - Reference and troubleshooting