1 StarOffice / OpenOffice
1.1 How do I get StarOffice to work?
1.2 (For SysAdmins) How do I get StarOffice to work over a network? (network install)
1.3 The printer doesn't print my OpenOffice documents. What's wrong?

2 Reading and writing to tape
2.1 How do I use the Exabyte/DAT drive?
2.2 What tapes can I use in the Exabyte/DAT drive?
2.3 Why is writing to the Exabyte drive so slow?
2.4 What type of drive is it?
2.5 Why do my Exabyte tapes get mangled?
2.6 How can I read a FITS tape (fits files written directly to tape, not using tar)

3 Backups and recovering lost data
3.1 Are backups kept?
3.2 What are backed up?
3.3 What are not backed up?
3.4 When are backups made?
3.5 To where are they backed up?
3.6 How do I recover lost material?

4 CD Drives
4.1 How do I read a CD-ROM?
4.2 How do I burn a CD?

5 IRAF
5.1 Why do I get an error about a BadMatch when I try to start ximtool?
5.2 Why do I get gibberish on the screen when I try to plot something?
5.3 How do I get IRAF up and running?

6 Printing
6.1 What printers are there?
6.2 How do I print to them?
6.3 How do I print a postscript file generated for (European or ISO-standard) A4 paper?
6.4 How do I print an encapuslated postscript (EPS) file to the poster printer?

7 Login between hosts
7.1 Why can't I log in using telnet/rlogin? Why do I get the error: "telnet: Unable to connect to remote host: Connection refused"?
7.2 How do I get SSH to log me in without asking for a password?

8 Misc
8.1 From where do the names of the computers come?
8.2 When I insert an XSPEC (or any PGplot) figure in a LaTeX file, it comes out sideways.
8.3 How do I do the correction for Chandra's CTI problem?
8.4 How do I split up MIME-encoded mail messages or files?
8.5 How do I change/setup my default window manager?

Change to your home directory:
cd ~

Run the setup programme. This will create a local subdirectory in you home space and copy some files to there.
/usr/local/OpenOffice/program/setup

You will get a GUI window which will guide you through the local installation.

After the installation is done, to run StarOffice, just execute the soffice command in the location it was placed. For example, if the local installation directory was ~/OpenOffice.org1.0.2, then type:
OpenOffice.org1.0.2/soffice

To make it easier to start in the future, you may wish to create and alias. I keep my aliases in my ~/.aliases file. So I have in that file the line:
alias soffice 'OpenOffice.org1.0.2/soffice &'

1. As root, from a terminal, cd to the directory where you downloaded so-5_2-ga-bin-linux-en.bin
2. Type ./so-5_2-ga-bin-linux-en.bin -extract (I'll get to the reasonin a moment.)
3. From this same directory type: ./setup -net
4. During installation, change the default install directory from /root/office52 to /usr/local/office52 (and let the installer create this directory). Ignore warnings about Java and database files.
5. From the directory you extracted so-5_2-ga-bin-linux-en.bin, copy setup.ins to /usr/local/office52/program. (The reason is that when you log in as any other user and try to run setup, you will get a message that setup can't find setup.ins and refuses to let you create a small office52 directory in your home directory. The only way to get this file is to extract the so-5_2-ga-bin-linux-en.bin file in the first place.)
6. Delete the extracted files from your download directory.(Don't delete the so-5_2-ga-bin-linux-en.bin file that you downloaded if needed later.)
7. When this install is done, logout of root and login as a normal user.
8. As user, from a terminal, type: cd /usr/local/office52/program (enter) then type: ./setup
9. Follow the prompts and you're on your way.

Change your page size to Letter.

You will need to do this for both the page layout, and the printer control.

Access the page layout via: Format -> Page -> Page -> Format .

For the printer control: File -> Print -> Properties -> Page .

There is no way (AFAIK) that a user can change their default page size. But the SysAdmin may do so by editting the file:
/path/to/OpenOffice/share/psprint/psprint.conf
In this file, look for the parameter: PPD_PageSize . Set it to an appropriate value, like Letter .

You will need an account on jca. Login in to jca.

For each drive, there are a multiple of logical devices you can write to that will write to the tape drive, but vary the capacity or level of compression.

Select the device (/dev/nrstxx) you wish to use and set your environment variable TAPE to that device. For example, to use the Exabyte drive, you can set the environment variable with the command:
setenv TAPE /dev/nrst20
After this, the standard tape control commands (mt and tar) will work normally, without a device explicitly specified.

Data class DAT/DDS and Exabyte tapes will work. For the Exabyte drive, look for 112M or 120M tapes (the 'M' means metres, not megabytes). The 112M may be more stable than the 120M. For the Dat drive, look for DDS tapes 60m in length.

Note that DAT, DDS, and 4mm (?) are synonyms, as are Exabyte and 8mm. Consequently, the DAT (Digital Audio Tapes) that you may purchase (if you can find them) to record digital audio are the same thing and will, in principal, work in a DDS (Digital Data Storage) system. (Is there and computer storage system that isn't digital?) The same goes for the 8mm hand-held video recorder tapes, which are much easier to find. They will work in the Exabyte drive. Just don't buy the cheap ones because the gum up the heads, so the myth goes.

8mm tapes come in a variety of lengths. The recommended length is 112M. I've used 120M in older Exabyte drives without any problems, but note that I haven't found 120M Data tapes, only Video. The 160M tapes will work, but are not recommended. According to Exabyte, they lead to damage to the media and drive. But then Exabyte suggest purchasing only Exabyte media. I recommend using them only for data sets that just won't fit in 4.5 Gb.

Writing to tape from an NFS-mounted directory can be substantially slower than writing from a natively-mounted directory ( 1/5 the speed ). I don't know why that is. If at all possible, move the file(s) to be written to a natively-mounted drive of jca before writing. But that may not always be easy, since the native drives are not very large.

A solution is to move the files to be written, to a data drive on ix (sage or hops). The NFS read connection between ix and jca is about 200 times faster (no exaggeration!) than when reading from pak, daneel, or strickland.

I am hoping to one day track down the source of this problem and fix it.

The Exabyte drive purports to be an Exabyte EXB-8500 8mm Helical Scan tape drive. See Exabyte's site for information on this (discontinued) product.

The specs are: DTR = 500 kb/s, search = 37.5 Mb/s, capacity 5 Mb uncompressed, 10 Mb compressed (112M tape).

You ejected the tape using the mt offline command.

Yes, this should work, but it consistently mangles the tapes (based on 6 tests: 3 using mt offline and 3 not).

Instead, use mt rewind and eject the tape using the button on the tape drive. This method did not mangle the tape in any of the 3 other tests.

Use the script (on jca): /usr/local/bin/fitstaperead.

Yes.

On all machines backups are kept of all home directories, system files, and data areas. The exception is mrkurtz which does not have a home directory (it is NFS-mounted from jca).

The areas that are not backed up are /tmp, and the backup drives.

There are three primary backup drives, /data/anise, /data/lovage, and /data/quinine.

All three are mounted on pak.

A drive mounted on doron, /data/ginseng, has an area devoted to backing up pak's home, system, and data drives.

Now is a bad time to ask yourself when you last went to church.

It can be done by yourself, but best to see Eric Tittley since there are a few details missing, particularly concerning the incremental backup system.

To get the CD in a position where it's contents may be read, it must be mounted.

1. Insert the CD in the CD drive
2. Type: mount /cdrom
   The CD is now mounted under the mount point /cdrom
3. e.g. cd /cdrom
4. e.g. ls /cdrom

To remove the CD after you are done with it, it must be unmounted.

1. Type: umount /cdrom
2. Eject the CD from the drive as you normally would

Burning a CD involves three steps:

1. Collect what you want put on the CD in a directory
2. Make an image file of that directory
3. Burn that image to the CD

Collect
Fairly straight forward. Simply create a directory and move/copy whatever you wish to burn into that directory. The directory can contain subdirectories etc.

Create Image

1. Find a location that contains at least the same amount of space as is occupied by the contents of the collected directory.
2. Issue the command
   mkisofs -r -o cd_image directory_you_wish_to_save/

Burn

1. Insert a blank CD-W or CD-RW in the CD-drive. Do not try to mount it!
2. Get the device number for the CD-drive.
   sudo cdrecord -scanbus
   This will produce an output like:

           
   Using libscg version 'schily-0.1'
   scsibus0:
    0,0,0  0) 'HP' 'CD-Writer+ 9100' '1.0c' Removable CD-ROM
    0,1,0  1) *
    0,2,0  2) *
    0,3,0  3) *
    0,4,0  4) *
    0,5,0  5) *
    0,6,0  6) *
    0,7,0  7) *
           
           

   From which it is seen that the CD-Writer is the only SCSI device connected and has been assigned the device number 0,0,0
   
3. Now burn the disk. Since Read-Write (RW) disks are a bit more complicated than a Write-Once (W) disk, the write speed is different.
   1. For a CD-W disk, issue the command:
      sudo cdrecord -v speed=8 dev=0,0,0 cd_image
      
   2. For a CD-RW disk, issue the command:
      sudo cdrecord -v speed=4 dev=0,0,0 cd_image
      

That's it! Now just mount the CD as you normally would and check its contents.

If you get an error when running the sudo cdrecord ... command stating that you do not have the permissions to issue that command, then contact the SysAdmin so that you may be given the right to write.

ximtool is an obsolete legacy tool. It cannot support colourmap environments with anything other than 8 bits per pixel. Most machines here, (with the ignoble exceptions of the Suns), are 24 bits per pixel (i.e., truecolour).

There are two work-arounds:
1) Use a modern image tool
2) Start your X-session in 8 bit mode

Solution 1):
Use ds9. Start it before IRAF is run via the command:
ds9 &
If the previous doesn't work, try:
ds9 -fifo /dev/imtl &
Better yet, alias ximtool to ds9 & in your .aliases file (assuming you have one)

Solution 2):
I don't want to go there.

Your terminal is the wrong type. Make sure you are using an xterm or an xgterm. Terminals of the type rxvt, while better for most every other purpose, don't work with IRAF.

To find out what type of terminal you are running, try:
ps -f -u $USER

Before running IRAF in any directory in which you intend to work, you should install a login.cl file by running:
mkiraf

There are two points of view on this. Either:

1. install the login.cl file in only one place, and always start IRAF there, then change your working directory within IRAF to where your data is.
2. install a login.cl file wherever you intend to work on your data, then change to that directory, then start IRAF.

IRAF itself is started by typing:
cl
which has to be the most unintuitive of the many unintuitive things in IRAF.

We have immediate access to 5 printers:

1. B/W HP laser in Rm 306A (can do double-sided printing)
   (For Toner replacement -- see Kathy/Cathy)
2. B/W HP laser in Rm 205 (can do double-sided printing)
3. Colour HP laser in Rm 205 (can do transparencies on a good day)
4. Colour poster inkjet printer in rm 212
5. Colour Tektronics printer in Rm 404E

Issue the following commands:

Note, the command lpr should be replaced by lp if you are on one of the Solaris boxes.

Also note: the poster printer is not available on the Solaris boxes.

The ISO standard A4 page is about 0.7 inches longer and 0.2 inches narrower than the American Letter standard page. A4 paper is in use everywhere except the US and Canada, so any postscript file you get from European or Asian collaborators or journals will be in this format. Printers set up for US letter paper will generally chop off the top of the a4 page, rendering about the first 0.5" of your file invisible. Obviously this is undesirable.

Adobe Acrobat Reader (acroread) has the capability to fit a page of any size into a given paper size. However, acroread cannot read postscript files.

What you have to do is first issue the command
ps2pdf -sPAPERSIZE=a4 myfile.ps

ps2pdf will convert any postscript file to pdf format (i.e. it will create myfile.pdf above), which is what acroread takes. The -sPAPERSIZE=a4 is essential - otherwise ps2pdf will cut off the top 1/2 inch of each page in exactly the same way as when printing it

After running ps2pdf, issue the command
acroread myfile.pdf

Now select the "print" command (either through the menu or icon bar) and print your document, making sure that the "fit to page" option is selected.

You're done! The document will be reduced in size by 6%, but it will be readable.

This problem can come up when users create posters with packages that produce EPS output instead of regular postscript. The problem is that EPS doesn't contain page-size information.

The solution is to run it through ghostscript to generate a PS file. This is done with a command similar to:
gs -sDEVICE=pswrite -sOutputFile=temp.ps -q -sPAPERSIZE=a0 -c quit FileYouWantPrinted.eps

Note to change the papersize to the size appropriate for the file. A0 and archE are good guesses.

Now print the file temp.ps with the command: lpr -Ppsposter temp.ps and wait 20 minutes.

You can only log in using secure shell (ssh) Version 1 or 2 on the machines:

• daneel
• doron
• hoth
• ix
• krupa
• lancelot
• mrkurtz
• pak
• strickland
• telescope
• zenas

If you are at a location that does not support ssh, then log in to your UMBC account on umbc7 or whatever. From there, you may use ssh to get into the JCA computers. Use the syntax:
ssh -l [your JCA account name] [machine]
if your JCA account name differs from your UMBC account name.

SSH allows for authorization with out the use of a password. It does so by comparing remote and local `keys'.

This can be very useful, and is also generally more secure since it means that passwords are not typed and hence a local snooping program can't get your password as you type it in.

It is also required to run remote applications that need to spawn local programs. AIPS, for example, uses this feature.

Getting it to work requires doing three things: 1) configure SSH; 2) configure your environment; and 3) load a passphrase.

Configure SSH

1. In your home directory, run:
   ssh-keygen
   
   • You will be prompted for a passphrase (password) which you need to remember.
   • This will create two files:
     either SSH
     $HOME/.ssh2/id_dsa_1024_a
     $HOME/.ssh2/id_dsa_1024_a.pub
     or openSSH
     $HOME/.ssh2/id_rsa
     $HOME/.ssh2/id_rsa.pub
     
2. Create the file $HOME/.ssh2/authorization with the contents:
   either SSH
   

   *:
   	Key	id_dsa_1024_a.pub
   

   Note that the spaces are tabs!
   or for openSSH, copy $HOME/.ssh2/id_rsa.pub to $HOME/.ssh2/authorization.
   `
3. Create the file $HOME/.ssh2/identification with the contents:

   *:
   	IdKey	id_dsa_1024_a
   

   Note that the spaces are tabs!
4. Copy the four files:
   • $HOME/.ssh2/authorization
   • $HOME/.ssh2/identification
   • $HOME/.ssh2/id_dsa_1024_a
   • $HOME/.ssh2/id_dsa_1024_a.pub
   to all your unique $HOME/.ssh2/ directories.
   Note that in most cases, your home directory is shared so this is not necessary. The Solaris boxes have their own shared home directories, so put the files there. And you can put them on your home and GSFC machines, too.

Configure your environment

ssh-agent must be running in a shell below the one in which you are working. The easiest way to do this is to have it run under the window manager. That way, every window in which you work can take advantage of the no-password login.

The problem is that everyone will have a different window manager and setup for the window manager. So I will use my setup as an example (fvwm2 started from ).

I have two files that are nearly identical in my home directory:
$HOME/.xinitrc
$HOME/.xsession
The difference is that one is used if you log into a console window and start x by executing startx and the other is used if you log in via a login manager (graphical interface). They are almost identical. The difference is that one needs to be a csh script and the other a bash script. (This might not be strictly true, but I've found these files to be finicky, and what I've settled on seems to work constently on all the JCA machines.)

These files are mostly filler. The main thing they do is start up your window manager, which is also the last thing they do. So here is what the last 5 lines look like in mine

if [ -d $HOME/.ssh2 ]; then
 exec ssh-agent fvwm2
else
 exec fvwm2
fi

if ( -d $HOME/.ssh2 ) then
 exec ssh-agent fvwm2 >& .fvwm/fvwm2.err
else
 exec fvwm2 >& .fvwm/fvwm2.err
endif

These won't take efect until the next time you log completely out of your console machine and log back in.

Load a passphrase

The next time you log into a machine at the console, open a terminal window and execute:
ssh-add
You will be prompted for a passphrase. Give the one you used when you created your keys (Step 1).

At this point, you should be able to ssh to any other machine, without needing a password.

IMPORTANT CAVEATE

Clearly, password-less logins are only more secure if you NEVER LEAVE YOUR COMPUTER UNGUARDED. Lock your door when you leave. If you are in a semi-private area, (e.g. 306A), lock the screen using xlock.

Run the correct_cti.pro routine.

This is an IDL code, so you will need to run it on daneel. Since IDL is a single-key license, only one IDL session can be run at a time. Eric Perlman is the one who paid for it, so it should default to him if a conflict arises.

The code makes use of PSU's TARA library of IDL routines. In your ~/idl/startup.pro file, add the paths as follows:
!path='/usr/local/share/idl/corrector: /usr/local/share/idl/astron: /usr/local/share/idl/astron/pro: /usr/local/share/idl/tara: /usr/local/share/idl/tara/event_browser: /usr/local/share/idl/tara/utilities: /usr/local/share/idl/tara/widget_tools:' + !path

1. log onto daneel.
2. change to the directory containing your my_obs_evt1.fits data.
3. start IDL with: idl.
4. at the IDL prompt, type
   correct_cti, 'my_data_evt1.fits', OUTFILE='my_data_evt1_CTI_corrected.fits'

For more details, see Leisa Townsley's notes.

The previous can be done from the command line by running:
/usr/local/share/idl/corrector/correctit my_data_evt1.fits

IMPORTANT CAVEAT: The examples above assume the CCD temperature was -120C. For other temperatures, a flag must be passed. See Leisa Townsley's notes.

Use mhstore.

If it is a mail file and your mail reader does not automatically break apart MIME-encoded files, then first save the file to something like myfile.mime, then separate the components with:
mhstore -f myfile.mime.

The answer depends on which machine you are working. Some run gdm for their login manager. Some use xdm. The Solaris boxes use dtlogin. For gdm, there will be an option on the login manager letting you select the window manager you wish. This may or may not work. I've found it to be a bit of a black art. Which is the reason I prefer to have xdm be the login manager. It is more primitive, but more predictable.

The solution that follows pertains to only the Linux-ix86 machines running xdm as their login manager.

In your home directory, create the file: .xsession with the following contents:

#!/bin/csh
set WM_CMD="ssh-agent fvwm2 >& .fvwm/fvwm2.err"

if ( -d $HOME/.ssh2 ) then
 exec ssh-agent $WM_CMD
else
 exec $WM_CMD
endif
   

Make the file executable:
chmod +x .xsession

The script will set up the window manager to be FVWM. To run Gnome, use the line:
set WM_CMD=gnome-session
For KDE:
set WM_CMD=kde

It might be a good idea to copy .xsession to .xinitrc in the event the login window does not appear. If you log in from the text console, you can start your X-session with: startx which looks for $HOME/.xinitrc.