Backups and File Recovery
Snapshots - File server appliance
The Department's Oracle file server appliance provides redundancy in terms of controllers of the file storage array, parity error correction on the file storage system itself and a complete mirroring of all the file systems on the appliance (each user has their own file system, or share, on the appliance, which has a mirror). This redundancy provides a robust backup for the members' of the Department home directories on an ongoing basis.
In addition, for the occasional "Oops" that a user may have, accidentally deleting a file or overwriting its contents, the file server appliance provides "snapshots" of changed files. There 8 daily snapshots (performed at about midnight of each day) and 5 weekly snapshots (performed at about 1am each Monday). These snapshots can be thought of as an incremental back-up, so only changed files within the given period are copied into a snapshot, although the snapshot can access all files within a file system.
*Note: Changed files are only captured after midnight on each day, so if a user creates a file, edits it and then deletes it, the user will not be able to recover that file unless the file was created before midnight and then deleted after midnight.
The snapshots are available in each user's home directory under the .zfs/snapshot directory.
- A typical snapshot directory will be named something like: .auto-1385971260
- You can tell which snapshot is the latest using the "ls -al" command in your .zfs/snapshot directory
Back-ups
System wide back-ups are performed on a daily basis. Of highest priority are back-ups of all home directories. If a file or directory is created and left in place for at least 24 hours our back-ups should have a copy of it. If a file does not exist for a minimum of 24 hours it is likely that our back-ups will not contain a copy of that file.
Although the Department of Computing Science attempts to provide and maintain an extremely reliable back-up service on its systems, students should not expect this service to be 100 per cent reliable. Hardware/software malfunctions and tape media problems can affect the availability of data from the back-up service. Always treat your data with care and attempt to maintain its integrity on your own.
File Recovery
If a user accidentally deletes a file or directory that has been in existence for more than 24 hours it is likely that the file can be recovered. Users can submit a request to Help Desk for a file to be recovered.
The message requesting the recovery should indicate the following:
- The exact path of the file/directory to be recovered
- The date from which the file should be recovered (a user might not want the most recent version)
File/directory recoveries will take, at least, a day to perform. The amount of time necessary to perform a recovery increases as the amount of data to be recovered increases.
If you delete a file an hour before you wanted to submit it for the submission deadline, it is extremely unlikely that it can be recovered prior to the submission deadline. In this case you should submit the request for the file recovery to Help Desk and to your instructor or TA (so they know what has happened).