xgridmanage is a script designed to create a more user friendly nterface to the OSX xgrid command line utility. I wrote this project to help out one of my professors and a research student of his.

NOTE Program is in need of upgrading.

This project (at the time of writing is hosted on github)

http://github.com/davidwisch/xgridmanage

xgridmanage does not need to be installed, however a simple script is provided that copies it into a directory named '~/path' and updates either your .bashrc file or .bash_profile file to include that directory in your PATH.

To run the setup script, cd into the same directory as it and run:

NOTE The script is terrible, it's the first bash script I ever wrote. I need to update it.

sh ./setup.sh

You can specify the xgrid server host and password via a configuration file. These options are specified in a file named .xgridmanagerc and should be placed in your home directory. The syntax for this file is as follows:

HOSTNAME localhost
PASSWORD mysecretpassword

A sample config file is provided with the rest of the files. The real config filename must be named .xgridmanagerc.

For a list of accepted commands run 'xgridmanage -h' or 'xgridmanage --help' or just 'xgridmanage' with no options at all!

When you submit a job to the xgrid, a file called submit_ids is created. This file is what stores the job ids for everything that was just deployed - don't delete it. This is the file you'll want to pass back to xgridmanage to retrieve all your results (and to delete those jobs from xgrid).

//Submit program.bin to 10 machines
xgridmanage -n 10 -d program.bin

//find status of all program.bin jobs
xgridmanage -s submit_ids

//retrieve results for program.bin
xgridmanage -b submit_ids

//delete all program.bin jobs
xgridmanage -f submit_ids

When you retrieve results from xgrid, they're placed in a folder named RESULTS. This folder contains two types of files, text files that contain the stdoutput for the jobs you retrieved and directories containing any output files those jobs may have been created.

The files containing the stdout are of the form "id_[job_id]_stdout". The folders containing output files are of the form id_[job_id].

Where in both of the above examples '[job_id]' is replaced by the actual job id.