What IS Backed Up?

Lots of stuff. Nearly all support-managed networked file systems are backed up. Generally speaking these categories include: all centrally hosted home, web, course, project directories and databases; as well as all configuration directories critical to the operation of file servers and our hosted services, e.g. "/etc" .

The CS Lab uses a primarily disk based backup system. The system used was designed in-house, leverages both rsync and the snapshot abilities inherent to the ZFS filesystem, and is referred to as ZBACKUP  (z is for ZFS).

The list of filesystems that are backed up with ZBACKUP  can be found here: Zbackup Filesystem List. The list of directories being backed up routinely changes.

What IS NOT Backed Up?

Lots of stuff. 


Some filesystems or directories are specifically excluded:

  • These filesystems are explicitly NOT Backed Up
  • Any UNIX directory named "nobackup"
  • Additional directories used by certain apps like Thunderbird and Firefox either to cache files or as a way station to oblivion are excluded, viz. "ImapMail", "Cache", and "Trash".

But what about Cloud Email?

CSE email provided through Google and Microsoft offer the standard vendor managed "30 day trash retention" to individual mailboxes. That is, when an email message is deleted it is moved a customer's Trash 'folder', and after 30 days those message are expunged from the service. Manual customer-initated deletion from Trash can speed up the deletion process. There is no way for a customer (or CSE IT staff) to recover the original email from a user mailbox once it has been deleted from trash.

File Restoration Requests

To request restoration of a missing or corrupted file, contact support. Include in your message the last date (and even the time of day) that the file is likely to have been last changed or written.

Backup Policy

Files are backed up from primary disk volumes to backup volumes for almost all CSE Lab file servers, as well as for a number of other computers. The primary purpose of backups is to allow recovery from loss of primary volumes. Recovery procedures are oriented mainly toward restoring all data on a volume, but specific files or directories may also be recovered. Backup retention policies vary based on the type of data involved, but generally are between 6-12 months.

Backups are not perfect. We mitigate some of these problems by vigilant monitoring of daily backup reports and by replicating data in multiple places, and finally by taking care that almost all backup volumes and their corresponding primary volumes are situated in different buildings. For example, all backups of file servers in the Allen Center are replicated on disk volumes in Gates Center and vice versa. 45-90 day "synthetic point in time full" snapshots of our filesystem are also securely transmitted to UW IT LOLO archive service for disaster recovery purposes.