Notes for ITC Staff

Unix Filesystem Backups at ITC

(this page last updated 2008/09/06)

Notes for ITC Staff

A user may request a restore of a specified list of files or directories. This is typically a set of files that a non-root user has accidentally removed or corrupted. The request should be made by telephone call to the ITC Help Desk. Help Desk staff can restore onto a local disk and then copy the files to a remote machine. Staff may obtain assistance from the Systems or Networks group if any of the restored files are owned by privileged accounts such as root, or require other special treatment.

A system administrator may request the restore of one or more entire filesystems. The request should be referred to the ITC Unix Systems group. The restore should not be handled by the user admin method (see below), for the following reasons.

The amount of data involved is usually large, which could exceed the capacity of local disk, or use up so much local disk that other restores are inhibited.
Once data is restored to local disk, this large amount of data must be transferred to a remote machine. The extra step delays the restore.
Some copy/archive utilities do not preserve holes in files and may create extra copies of files if extra hard links exist. So unless the transfer is performed using the uvadump/uvarest commands, there is a risk that the final product may not be a faithful recreation of the original filesystem, and that expanded holes and extra copies could prevent a successful transfer because of lack of space in the destination filesystem, even if the size of the filesystem is the same as the original.
Filesystems may contain files owned by root, which would have to be transferred by Systems or Networks staff.
Scans of the backup archive catalog to detect files which were in use at the time of the backup are not done. The information can be displayed with browsing commands during an interactive restore, but it is unlikely that anyone attempting to restore an entire filesystem interactively would go to the trouble of displaying information for all files in the filesystem, looking for in-use files. The result would be failure to detect unusable files or files in the wrong directories.

Systems staff should use the /common/ud/a/u/access script to perform a filesystem restore.

Note:

In this document, publicly accessible on the Web, the fictitious name T.itc represents either the name of the machine handling a robotic tape library or the name of the machine handling virtual tape libraries. A.itc and H.itc represent the names of the administrative machine and the onsite backup storage machine, respectively. The corresponding actual names, which may be found in /common/ud/a/u/access, are not used here to avoid inviting intrusion attempts.

File restores may be done by logging into one of two machines as user admin. Use machine A.itc to restore an email inbox; use machine T.itc for other restores. Select the br option. Select the machine and filesystem to restore from. For either of these, it often is not necessary to type the entire name; just type enough of the name to make it uniquely recognizable. Select an approximate date for the archive to look for. Select an archive. Mark the files or directories to restore. Type "g" when done; at this point, you will see a message starting with ":d253 Archive st". On T.itc a delay of a few minutes should be expected while the robot accesses the appropriate tape. After the restore finishes, the "co" menu option may be used to change the ownership of user-owned files within the restored directories. You may want to take a look at sample sessions to do a restore or to change ownership of restored files.

It is sometimes necessary to track down the location of a file. For example, suppose a user requests the restore of the file /www/data/D/F on machine M. You may not see this path to the file when browsing through a backup catalog because of the presence of symbolic links, filesystem mounts, or pseudo-filesystem definitions. To find the "real path" to F, login to M as userv. cd into the directory containing the file:


  cd /www/data/D
  df .

df tells you the filesystem containing D (which contains F). Also type


  /bin/pwd

to find the real path to D. This is the path you will have to follow during browsing.

Finding the real path may not be possible if userv does not have permissions to traverse the path on M, or if M is not up, or if the path no longer exists on M. In these cases, the real path can still be discovered by following the path given by the user one element at a time during browsing. For the above example, this would mean browsing a backup catalog of the / filesystem to find "www", then "data", etc. You may have to browse more than one backup catalog if a path element takes you to another filesystem.

Administration

Administration of the ITC Unix filesystem backups is handled by the ITC Unix Systems group.

Special Requests

Occasionally backups for a particular user's files must be preserved for legal investigation rather than having the backups removed as they expire. See comments in /common/ud/a/u/restfra for the procedure.