The Zoo cluster employs redundant arrays of independent disks (RAID) storage that isolates us from most device failures. Nonetheless, device failure can cause data loss and we all delete files we want back later.
Zoo files are auttomatically backed up every day shortly after midnight to a robotic tape library. Every file is backed up on a weekly basis. New and changed files are backed up daily.
The robotic library cannot hold all backups, but will typically hold all files backed up in the prior month or so. [This will vary with activity/disk utilization and increase as we add capacity to the online system.] You can check to see what versions of your file are still online using the file recovery facilities described below. Library overflow tapes are moved to an off-site storage area. Once files are removed from the robotic library, file recovery is much more difficult and time consuming. Should your files be at the offsite vault, technical assistance and a day or more may be required. Note that the back-up system is not intended for archival storage. For that purpose, you will likely want to choose your own media such as a CD-RW or other portable media.
You can easily restore your own files as long as they are still in the robotic tape library. While the instructions may seem a little complex at first, but if you have a basic familiarity with the unix login environment you should have no trouble. Give it a try. If you run into trouble, contact the helpline@uvm.edu.
To use the "recover" command, you have to log into your zoo account.
The recover command offers a command line interface for restoring files. Entering recover at the unix prompt will bring you to the recover environment (shell). The recover shell is much like being in the unix shell, in that many of the everyday unix commands, such as "ls" and "cd" work.
A quick example:
This morning MGA deleted a file by mistake, let's call it "really.important.file". The following is what the recovery process for that file looks like. Note that user commands are shown in blue.
|
|
|
|
mga@zoo.uvm.edu> pwd |
Command prompt for user "mga"; pwd = print working directory |
|
/.../uvm.edu/fs/racka2/tsg/mga/mail
|
Output of pwd command, i.e. the current working directory. Note that the current directory is the mail directory. User mga must have issued a "cd mail" command prior to pwd... |
|
mga@zoo.uvm.edu> rm really.important.file |
remove really.important. file; [don't try this at home :-)] |
|
rm: remove really.important.file? y |
remove confirmation; your second chance. |
|
mga@zoo.uvm.edu> ls -l |
ls -l = list files in current directory with "long" option (i.e. with protections, ownership & data stamp) |
|
total 24 rw------- 1 mga system 3309 Sep 20 13:19 saved-messages -rw------- 1 mga system 3793 Sep 20 13:19 sent-mail
|
Output from directory list command (ls). Only 2 files listed. Total 24 is the amount of space consumed. |
|
mga@zoo.uvm.edu> recover |
recover command = enters the recovery environment so we can recover one or more files. |
|
recover: Current working directory is /.../uvm.edu/fs/mailhome/m/g/mga/mail
|
recover command tells us the current working directory (in case we forgot) |
|
recover> ls -l |
recover> prompt followed by ls command. Note that it is just like the regular unix ls command and does the same thing except that it shows you the most recently backed up files on the backup library server. |
|
total 2716 -rw------- 1 mga 3309 Sep 20 13:19 saved-messages -rw------- 1 mga 3793 Sep 20 13:19 sent-mail -rw------- 1 mga 2775552 Oct 28 11:30 really.important.file
|
Output of the recover command. Note that the total space is now 2716. That's because the "really.important.file" was pretty big (2775552 bytes). |
|
recover> add really.important.file |
add = add this file to my list of tiles I would really love to have back. |
|
1 file(s) marked for recovery |
Response from recovery. |
|
recover> list |
list = show me the list of files you are set to recover |
|
/.../uvm.edu/fs/mailhome/m/g/mga/mail/really.important.file @ Sun Dec 1 06:25:35 1996 1 file(s) marked for recovery
|
Output of recover list command. Note that this is differnt from the ls command. |
|
recover> recover |
The recover> recover command -- starts the recovery process |
|
Recovering 1 file into its original location Total estimated disk space needed for recover is 2.7 MB. Volumes needed (all on-line): UVM.1168 at /dev/rmt0.1 Requesting 1 file, this may take a while... ./really.important.file Received 1 file from NSR server `ozzie.uvm.edu'
|
Output from the file recovery process. Upon completion (usually a minute or two), your file will be back at its original location. |
|
recover> exit |
exit = leave the recovery environment |
|
mga@zoo.uvm.edu> |
Back at the zoo command prompt. You can proceed to whatever you wanted to do with the file or just logout... |
Commonly used commands in recover:
|
|
|
|
|
Enter the recover program, or actually recover the selected files. |
|
|
Directory listing of all files in the online index for the current directory. |
|
|
Adds the filename to the list of files to be recovered. |
|
|
Displays what files are marked for recovery. |
|
|
Quits out of the recover program. |