Kepler

Setup

These are instructions for installing the Kepler release for data version 4.10a

Install a private version of KEPLER for 4.10a on your account:

cd ~
rm -rf kepler .kepler
tar xvf $SWITMDIR/kepler/4.10a/kepler.tar

Set the Kepler, UAL and database environment variables. Kepler will work in your private database, under the machine TokamakName (TokamakName=test is allowed for testing purposes).

source $ITMSCRIPTDIR/ITMv1 KeplerFolder TokamakName DataVersion

KeplerFolder should be your private Kepler path relative to your home directory. Simply "kepler" is recommended, unless you want to have several private versions of Kepler in your directory. Example:

source $ITMSCRIPTDIR/ITMv1 kepler test 4.10a

If this is not run on restart, kepler will open in the default environment which is not what you want!

To have your environment set by default (mandatory for new features such as debugging, use of batch jobs, ...), it is recommended to set default envinronment variables by adding this command in you .cshrc file as follows: source $ITMSCRIPTDIR/ITMv1 KeplerFolder TokamakName DataVersion >/dev/null

Prepare the input data (see Databases for details). Create a "test" zone in your private ITM data tree if you haven't already done it.

$ITMSCRIPTDIR/create_user_itm_dir test 4.10a

This version of Kepler is provided with a few workflow examples, located in the kepler/demos directory. To use them, you need to copy the example input data set (shot 4, run 1) from the public database to your test database.

cp /pfs/itmdb/itm_trees/public/test/4.10a/mdsplus/0/euitm_40001.* \
   $HOME/public/itmdb/itm_trees/test/4.10a/mdsplus/0

Run KEPLER:

kepler

Select File > Open File and open any of the example workflows in the demos directory. Start for instance with staticexample.xml. Results will be written to your private database.

ls -gtr $HOME/public/itmdb/itm_trees/test/4.10a/mdsplus/0

Issues

1. If the kepler environment variables are not set as described above, kepler will open in the default environment which is probably not what you want. Kepler will probably complain about missing files. Check that $KEPLER points at your private version.
2. If Kepler refuses to run. it may be because you have last run it on a different gateway node. Delete .kepler and try again!
3. If Kepler dies, the core dump can be enormous - 1.6GB seen with an ITM test workflow. To find core files in your home directory tree do

   find ~/ -name 'core.*'

   and to check disk quota, do

   fs listquota

4. If you have selected the debug mode of an actor (configure actor), Kepler will fire up totalview when running this actor. The process which starts totalview will do so in a new shell. Therefore, any .cshrc or .login file will be sourced again and environment variables (like TOKAMAKNAME or UAL) will be reset to the values defined in these files (if you included for instance a source of the ITMv1 script). This may result in Kepler not being able to access the runwork data file anymore since this was stored at the original location in the data base.
   If running actors in debug mode, make sure that TOKAMAKNAME and UAL set in .cshrc and .login are what you expect them to be!