After an Experiment

Experiment Closeout

After your beam time is over, please make sure you follow these steps before you leave:

1. Return Beam Blocker Key to the Cyclotron Operator in Charge.

The beam blocker key is provided by the operations group for the experiment running. This key must be inserted into the beam blocker panel located in the electronics rack in each data-U to actuate the beam blocker and to take control of the beam. This key is also needed to enable the trigger clock reset pulse.

Since the experiment running needs to be in possession of this key and since there is only one key, it must be returned to the Cyclotron Operator in Charge immediately after your beam time is over.

2. Closing the Experimental Account

It is the experimenters' responsibility to make backup copies of the data, configuration information, documents, etc. within two weeks after the beam time for the experiment is over. Details of how to do this are posted at:

http://docs.nscl.msu.edu/daq/accountclosing/accountclosing.pdf

Once you have done this, the Spokesperson of the experiment should notify Raman Anantaraman that the account's files have been properly archived. We recommend that for outside user experiments, two copies of the data, etc. be made, one of which is left at NSCL with NSCL collaborators or with Raman Anantaraman. The computer group will provide tape drives, CD burners, and DVD burners for this purpose. Please be sure to use supported media (LTO tape is preferred) if you expect to be able to read your backup tapes at NSCL. After Raman has been notified, the computer account will be closed and files deleted. Once an account has been deleted, it cannot be recovered by the computer group.

If the experiment will be run in two or more separate segments, the experimental account can be kept open till the last segment is completed. However, it is expected that this account will not be used for data analysis in the periods between the run segments.

It is the experimenters' responsibility to ensure that the data and files associated with an experiment are retained and maintained in appropriate storage areas, as required for data analysis and other needs.

3. Leaving Vault and Data-U Clean

After the experiment is complete, please make sure that the experimental vault and the Data-U are left in a tidy state. Outside users should allow enough time for this before they leave.

Preparing for Data Analysis

1. What to Expect on a Data Tape and How to Retrieve a Run

The stager and run control GUI maintain a directory structure which packages all information associated with a run into a directory tree. The directory tree associated with a run contains not only the event data file for that run but also any data the experimentalists associate with that run by placing under the ~/experiment/current directory tree by the time the run ends.

The directory trees staged in a given pass of the stager are put on tape in a Unix tar (Tape Archive) file. The directory structure of the tar file is described in:

http://docs.nscl.msu.edu

Click the overview link and then click the cylinder labelled "Event Logs" on the block diagram that brings up.

To recover a set of runs from tape, first look at the TapeIndex.txt file in your ~/experiment directory and then mount the correct tape in the drive. You may need to space forward a few files to position the tape drive at the file containing your runs. If the name of the non-rewinding tape drive special file is in the environment variable TAPE (e.g. on Tru64 Unix /dev/ntape/tape0) type:

mt —f $TAPE fsf n

where n is the number of files you need to skip to get to the file containing the desired runs. If you want to recover runs m,n,o and p, type:

tar xf $TAPE runm runn runo runp

The data will be recovered into subdirectories named runm runn runo and runp below your current working directory. These subdirectories will contain the event data file and files associated with the run while it was in progress

Note: there are collaborations at NSCL that do not use the stager software. Consult with them about how to figure out which runs are on which tapes.

2. How Access to Data is Managed after the Experimental Account is Deleted

Once an experimental account is deleted, analysis must be performed either on the NSCL general purpose compute systems or, ideally, on the analysis cluster. Regardless of where data analysis is done, the experiment must be provided disk space into which data taken online can be retrieved, and products of analysis placed. Disk space at NSCL for data analysis is divided into file systems intended for event data, which is considered to be read-only, and project directories. Request both of these by sending mail to helpme@NOSPAM.nscl.msu.edu
(remove NOSPAM from the address to get the real email).

3. Using the Analysis Cluster

The analysis 'cluster' is a load leveled set of linux systems. Each system has an identical hardware and software configuration. At present, access to the cluster is via ssh to spice.nscl.msu.edu.

3.1 Accessing Tape Drives

Tape drives are located on the the tapehost.nscl.msu.edu system. These drives are labelled (e.g. st0).
The labels refer to the name of the device special file corresponding to each tape drive in the /dev directory (e.g. /dev/st0 is labeled st0 and, /dev/st1 is labeled st1). Note that several device special files represent each tape drive. Special files whose names begin with the letter ‘n’ do not rewind when the program using them exits. Special files whose names begin with the letter 's' do rewind on exit.

Typically you will need to recover data from tape archive (tar) tapes created by the data acquisition system. The following tar command shows how to recover e.g. run71 from nst0 to the current directory, assuming that the tape index has told you the run is on the 3’d tar file on tape:

ssh tapehost mt -f /dev/nst0 fsf 2 (Space forward to 3'rd file from 1'st);
tar xf tapehost:/dev/nst0 run71 (Read run from tar file at that point). <>

Feedback

At the end of your run, please complete the run summary feeback form.