Spigot Use: To insure everyone who needs access to the spigot computers will have it, we need everyone to adhere to the following rules:

• Priority for use of the spigot2 and spigot computers are as follows:
  • The current observer
  • The person transferring data to external drives
  • Any maintenance or testing of the spigot system by GBT staff or David Kaplan
• As any use of the spigot2 computer during a data taking session could readily result in the observations failing, during an observing session, only the observer (and his/her Co-Is) can use the spigot2 computer. This means the observer has the right to kill any processes running on the spigot2 computer once his/her set-up time begins.
• Once an observing session is over, the observer must immediately begin copying their data onto external disk. The only exceptions to this rule are (i) if another observing session is starting, or (ii) if previous arrangements have been made for a different group to transfer their data to external disk.
• Once his/her data has been copied to external disk, the observer must delete their data from the spigot and spigot2 computers. If you do not do this, the next observer may be unable to take data.
• All test data must be moved off the raid disks before the next observing session begins (if it is necessary to keep the data, move it to the /home area).
• A listing of use of the spigot computers is available online at http://http://www.gb.nrao.edu/gbt/spectrometer/spigot_card/comp_use.shtml. (For now, you must send an email to Karen O'Neil, or use the resource calender in order to sign up for time. This will be made more flexible soon.)

Copying Spigot Data to an External Disk:

First, be aware that for fastest speeds you should run no data reduction or other programs while you are copying data.

If you are using Firewire ports (by far the fastest option!):

On the spigot2 computer, type "source ~pulsar/scripts/spigot.sh" or "source ~pulsar/scripts/spigot.sch".
• Unmount any external disks currently mounted on the system:
  • First, type "df" and look for anything called "/mnt/XXX" where XXX="ehd", "ehd2", etc.
  • If there are no such entries, then no external disks are mounted and you are fine.
  • If there are any such entries, unmount them using the command "sudo umount /mnt/ehd1" (or ehd2 or whatever). Repeat this until there are no more /mnt entries.
• Unplug any external disks you do not wish to have plugged in, and plug in (and turn on) those external disks you wish to use
• Type "sudo rescan-scsi-bus.sh -r" and wait until the program is done
• You now need to mount the new disks with a command along the lines of "sudo mount /dev/sdX1 /mnt/ehdX". The letter replacing the X in /dev/sdX1 ranges from g - ?. To determine the correct letter, run the command "sudo sg_map -i -x" and look for the recently plugged in disk(s). Also, the X after "ehd" can be 1, 2, or 3. Please be aware, through, that you can only have one drive attached to one mount point ("ehd" number).
• You can now copy your data onto the /mnt/ehdX disk using the rsync command:
  "rsync -avxuP /data1/observer/koneil /mnt/ehdX/mydata"

  If you are using USB2 ports: To copy data to an external disk, you first must unmount any disks using the port you need. To do this, simply type "umount /mnt/ehdX", where X=1, 2, or 3. Next, plug your disk into the appropriate port and mount your disk with "mount /mnt/ehdX". If your disk is the first to be mounted, X=1. If its the 2nd, X=2 and if its the third, X=3. Having done this, you should now be able to write to your disk (called /mnt/ehdX).

  Labeling your hardware: To insure your disks are readily locatable, and that all your various plugs and cords are returned to you, please clearly label your external drives and all cords with your name and a unique number. If you don't do this, we may not be able to guarantee the return of all your hardware.

  PROBLEMS/QUESTIONS

• I need to get my data off the spigot2 computer, but I can't find enough free time on the computer to do so. First, be sure that you are runingn no data reduction or other programs while you are copying data. that slows down the copy, and there is very little time to spare this month. Assuming you aren't, you again have two options. First, talk to the other folks using the spigot2 and re-arrange everyone's schedule for writing their data so that it accommodates you needs. If that doesn't work, grab a much smaller chunk of time (e.g. if you have only 4 hours after a 4 hour run) and instead move your data from spigot2 to spigot. To do this and avoid the very slow NFS connection, rsync calling the computers newspigot (a.k.a spigot2) and oldspigot (a.k.a. spigot).

• I'm scheduled to observe in 10 hours and I don't think there will be enough disk space for me! Before panicking, talk to whomever is observing or transferring data and find out how much data that person will be removing from spigot2. If that isn't sufficient (it should be!) then ask around and find find someone with data in spigot2 and ask them to please at least move it over to spigot and start copying it from external disk there.

• How should I copy data from disk to disk? The most reliable method is using rsync. the command would look like:
  "rsync -avxuP /data1/observer/koneil /mnt/ehd1/scratch/koneil" (for copying disk to disk) "rsync -av xuPnewspigot::data1/observer/koneil oldspigot::raid1/observer/koneil" (for copying from one computer to the next)

• How do I know what kernel is running on a computer? Type "uname -r"

• Which kernel should I run for data taking in spigot2? the spigot datataking system requires you to run the 24aa1 kernel

• I've done everything you've suggested, and I _still_ can't get my disk to mount. If it is the middle of the night, and it is possible, have someone else start transferring their data until morning, and then try calling Chris Clark or Wolfgang Baudler for help. If you have to deal with it now, try calling them anyway. To find out if Chris or Wolfgang are in town, take a look at https://wiki.nrao.edu/bin/view/GB/ComputingDivision

File last modified: Wednesday, 30-Jun-2004 17:51:54 EDT