NOTE: Please do not bring blank CDs or DVDs to Keck to use with the autoloaders. In particular, using the wrong type of DVD can actually burn out the drive! On the other hand, please DO bring jewel cases, CD/DVD sleeves, or some other means of protecting your disks when you take them home.

We currently have two autoloaders at Keck.These machines hold up to 120 blank disks, either CD-R or DVD-R disks, and can automatically burn and label data onto one or more of these disks. We have recently upgraded the autoloader in RemoteOps 2 to a dual-burner. It is currently undergoing testing, and will eventually become the default device for observers. In the meantime observers must share the autoloader in RemoteOps 1, on awalua, with the Keck data backup system (called "savier").

To make life simpler for observers, a tool called "autobackup" was developed. There are two main components: a tcl/tk GUI, and a csh shell script. The shell script is rather general, but the GUI is targeted directly at observers wishing to back up raw Keck data from their "/sdata" disk or disks. If you want to use the shell script, you can skip or gloss over the GUI description and head straight for the section on autobackup.csh.

Autobackup.tcl

The autobackup.tcl GUI is shown below.

Upon startup, the GUI checks your user name and looks for the INSTRUMENT environment variable (envar). For example, if you are logged on as "esieng", that would be your user name and "esi" would be the value of the INSTRUMENT envar. The GUI then searches the ESI data disks for directories owned by "esieng." (Specifically, it looks for directories of the form "/sdata*/esieng/*".) It then goes through this list and discards any that are empty. Note that in the "esieng" example shown in the figure, two directories were found.

The GUI then presents a list of the directories it has found, with check boxes next to each one. After each directory is an estimate of the amount of data in that directory (calculated using the "du" command). Note that you may have some older directories as well as your own. These are probably leftovers from a previous observer. In the case of instruments that are run from a single master account (NIRC and NIRSPEC), you will likely see a lot of directories. Note that one of the directories above says it contains "0 MB". This directory contains only a couple of subdirectories, and no real data.

Below the directory list are some options and information:

• Number of copies. This will produce 1, 2, or 3 copies of each disk. The disks will be labeled identically.
• Compression. Even on binary FITS files, the "gzip" compression routine can substantially reduce the size of the files. You may select this option if you are trying to fit large data sets onto the smallest possible number of disks. Realize, however, that compression will take some significant amount of extra time.
• Total space. This gives you the total space used by all of the directories you have selected above. If you have compression selected, this number is only an estimate (simply half of the uncompressed space).
• Type of disk. A CD will hold 650 MB of data, while a DVD-R disk will hold up to 4.5 GB. (We are somewhat conservative about the disk capacities to avoid disk overruns.) Make sure you have an appropriate DVD reader at home if you choose DVDs!
• Text for disk label. Four lines of text can be put on the disk(s) as a label. By default three of these lines are already filled in, but you can change any of the text. The size of the text boxes is a rough indication of how much text will fit onto the disk. You can put in more text, but some of it may end up off the edge of the disk. The text will be centered on the disk, not left-justified as in the GUI boxes.

Note that if your data will not fit onto one disk, it will automatically be partitioned into disk-sized chunks, and each disk will have another line in the label, saying, e.g., "Disk 1 of 5". Note that this is not "Copy 1"! If you are making three copies of data that fits on five disks, there will be three identical disks reading "Disk 1 or 5", three more labeled "Disk 2 of 5", etc.

Once you have selected your directories and options, click on "Write" and the GUI will pass the information on to the shell script, autobackup.csh. In principle, you should be able to go to sleep or do other work while the computer and the autoloader do their thing.

Autobackup.csh

This shell script is called in the form:

autobackup.csh dir1 dir2 dir3 [-n #] [-dvd] [-gzip] [-labelFile filename]

In general these options can be in any order, as long as the values of the "-n" and "-labelFile" options follow these options immediately.

The parameters are:

• dir1 dir2 dir3 ...
  A list of directories (full pathnames highly recommended!) to write to disk. Everything within these directories, and nothing outside them, will be written. Note that if more than one directory is requested, the program will have to copy all of the directories into a single umbrella directory. While this takes more time, it should be transparent to the user.
• -n #
  Specifies the number of coopies to make. There is in principle no limit to this, although of course if you exceed the number of blank disks currently loaded in the autoloader you can expect problems! The autobackup.tcl GUI currently only has options for 1, 2, or 3 copies.
• -dvd
  Specifies writing to DVD-R disks.
• -gzip
  Specifies compression using the fastest gzip compression. (Tests on LRIS-B and DEIMOS images showed that higher levels of compression took longer but made almost no difference in the final file sizes.)
• -labelFile filename
  Specifies a file containing text to incorporate into the label. When using autobackup.tcl this text is the same as the text in the boxes at the bottom of the GUI.

With these parameters, autobackup.csh will perform the following steps:

1. If no label file has been specified, a default one is provided.
2. If there is either more than one directory, or you are using compression, a scratch disk will be found and a staging directory created.
3. If you have requested compression, the data files are compressed into the staging directory. Any subdirectory structure is retained.
4. The data path and options are then sent to the command "data2auto2," which will partition the data into disk-sized chunks if neceesary, and burn and label the disks.
5. Any temporary staging directories and files are cleaned up.

Getting your disks

When autobackup.csh has finished, it will pause with a message telling you it is done, and waiting for you to hit <RETURN>. If you were using the autobackup.tcl GUI, the GUI will pop up again and allow you to burn more disks, change options, etc. If you are done with the GUI, hitting QUIT will kill it.

The disks can be found in either bin 2 or bin 3 of the autoloader. Feel free to rotate the black carousel containing the disks with your hand, but be extremely careful. If the autoloader tries to access the carousel while you are touching it, the carousel may spin around and cause some injury. You may have to pull out a stack of recently written disks to find yours. Other users may also have been backing up to the autoloader. Please handle all disks carefully, and return any disks you don't own to the carousel or to a CD/DVD spindle next to the autoloader. Conversely, if you see a spindle next to the autoloader, you should check there first for your disks. Someone else may have already gone through the output bin(s).