Backup names¶

Every backup set has a name. The name identifies the set in the logs and groups its files. ezbak adds a timestamp to each file, so each backup is unique.

The filename format¶

A backup file is named {name}-{timestamp}.tgz:

my-documents-20241215T143022.tgz
database-backup-20241215T020000.tgz

The timestamp uses the format YYYYMMDDTHHMMSS: a four-digit year, a two-digit month and day, the letter T, then two-digit hours, minutes, and seconds. The list command prints this exact timestamp for each backup. You can pass it directly to a point-in-time restore.

The name groups a backup set¶

ezbak matches backups by name, so several backup sets can share one storage location without a collision. A prune or a restore for my-documents never touches the files named for database-backup.

graph TD
  S["/backups"]
  S --> A["my-documents-20241215T143022.tgz"]
  S --> B["my-documents-20241216T143022.tgz"]
  S --> C["database-backup-20241215T020000.tgz"]
  A -.set.- B

Set the same name on each cooperating task, so they operate on one set. In the orchestration pattern, the sidecar, post-stop, and pre-start tasks share a name. The pre-start restore then finds the backups the other two wrote.

Timestamps and the timezone¶

The timestamp records the moment ezbak created the backup, in the configured timezone. In a container, set the timezone with the TZ environment variable. To override the system timezone, set the tz option (EZBAK_TZ) instead. When neither is set, ezbak uses the system timezone of the host. See Environment variables.